Browse Latest AI Freelance Projects from Upwork

Project Overview:
We are government contractors bidding on the Indian GeM (Government e-Marketplace) portal across 5–6 different registered entities (proprietary/partnership/LLP).
We need a qualified developer/automation engineer to build a complete 2-stage system:
1. Automated Eligibility Checker: Instantly screen an uploaded GeM tender against all 5–6 firms and output an eligibility scorecard (Eligible / Not Eligible / Conditionally Eligible via MSME relaxation).
2. Multi-Firm Document Generator: Automatically generate customized, letterhead-ready PDF annexures and compliance bundles only for the qualifying firms.
Key Deliverables:
⚬ 1. GeM Tender Extraction & OCR:
⚬ Parse GeM bid PDFs (e.g., GEM/202X/B/...), including image-based buyer-added text (ATC).
⚬ Extract key qualification parameters: Minimum 3-year Average Turnover, Past Experience / Single Order Value, EMD / Bid Security, MSE / Startup relaxation allowance (Turnover/Experience waiver: Yes/No), and Class-I MII % requirement.
⚬ 2. Firm Eligibility Screener (Firm vs. Tender Logic):
⚬ Evaluate extracted tender rules against our master firm database (Turnover figures, past work orders, Udyam MSME status, and blacklisting records).
⚬ Display a simple comparison table showing which of the 5–6 firms meet the criteria:
⚬ Firm A: Eligible (Directly meets turnover & experience).
⚬ Firm B: Eligible (Via MSE turnover/experience waiver clause).
⚬ Firm C: Ineligible (Turnover below threshold, no MSE exemption given by buyer).
⚬ 3. Master Database & Document Pipeline:
⚬ Build a central database (Airtable / Google Sheets / Supabase) storing static data for all 6 firms (GSTIN, PAN, Udyam, address, signatory names, past PO records, and high-res letterhead backgrounds with transparent seal/signature PNGs).
⚬ Auto-fill GeM-specific declarations (MII, Land Border Rule 144(xi), Non-Blacklisting, Turnover Undertaking, ATC Compliance sheets) into clean, ready-to-upload PDF packages customized for each qualifying firm.
⚬ 4. Distinct Metadata: Ensure generated PDFs carry clean, independent document properties to prevent administrative linking.
Required Skills:
⚬ Experience with workflow automation (n8n, Make.com, or custom Python FastAPI/Flask backend).
⚬ Proficiency with multimodal LLMs (GPT-4o, Claude 3.5 Sonnet, or Gemini 1.5 Pro) for OCR document extraction.
⚬ Experience with document templating engines (Docupilot, PDFMonkey, Weasyprint, or HTML-to-PDF).
⚬ Familiarity with Indian public procurement / GeM bidding rules (MSE turnover exemptions, ATC, MII) is a significant advantage.
Scope & Budget:
⚬ Project Type: Fixed Price / Milestone-based
⚬ Budget Range: Upto Rs. 25,000
⚬ Timeline: 2 to 3 weeks
How to Apply:
To confirm you read the full brief, please start your proposal with the word "GEM-ELIGIBILITY" at the top.
In your application, briefly state:
1. Your proposed architecture (e.g., n8n vs. Python backend, and which database/PDF engine you'd use).
2. How you plan to structure the rule-checking logic between the tender requirements and firm profiles.
3. Examples of similar document generation or parsing projects you've completed.
4. Confirmation that you are comfortable communicating in Hindi on calls.
for example file attached

We're a US-based wellness device company looking for an experienced regulatory and advertising compliance reviewer. We need a full audit of our Shopify website to find any language, images, or content that doesn't fit FDA's general 2026 wellness framework.

Scope

One Shopify site, an estimated 40+ URLs: product pages, FAQ/About/info pages, and policies.
Every place content lives: body copy, headlines, text inside images and banners, accordions/tabs/drop-downs, pop-ups, footer, reviews and testimonials, meta titles and descriptions, image alt text, product names, video, downloadable PDFs, and both desktop and mobile views (some sections differ by device).
Linked-out content, such as research or third-party pages linked from the site, since linked content can be treated as part of the promotion.

What you're looking for

Disease, treatment, or medical claims, explicit or implied, under FDA's general wellness guidance (updated January 2026).
Physiological or cellular-level claims, "healing" or anti-aging language, and implied claims in product names
Unsubstantiated performance, superiority, "backed by research," or "clinically proven" claims (FTC substantiation)
Testimonials and reviews with health and disease claims.

Anything else a regulator would circle

Deliverable
A spreadsheet listing with detailed notes

the URL
the exact text, or a screenshot for images
where it sits on the page when possible
a risk level (High/Medium/Low)
why it's a problem

Your role: identify and recommend. We make the edits, so no admin access is needed. Our legal counsel will review the final findings, so this is not a request for legal sign-off.

In your proposal, please include:

A quote (fixed)
Turnaround time
Your earliest start date. We need this ASAP.
Your methodology: how you'll make sure nothing is missed (images, hidden sections, mobile), and what might fall outside your reach
Your confidence level in catching everything that doesn't belong, and why
Relevant experience, such as FDA device or supplement promotional review.

Thank you!

Overview

I'm hiring a detail-oriented job application assistant to help me apply to full-time roles across San Francisco and the greater Bay Area. My background is in sales, marketing, and CRM, and I'm targeting early-career roles (listed below). You'll find relevant openings, tailor my materials to each one, submit applications on my behalf, and keep everything organized in a shared tracker.
This is an ongoing engagement with room to continue based on results. I also reward outcomes: if your work leads to a job offer I accept, you'll earn a sizeable bonus on top of your base rate (details under Budget & timeline).
About the candidate
• Early-career sales & marketing professional based in San Francisco.
• 2.5 years hands-on experience in CRM management (Salesforce, HubSpot), lead generation, outbound prospecting, and multi-channel marketing campaigns.
• B.S. in Management Information Systems, San Jose State University.
• Comfortable with analytics/reporting tools (Excel, SQL, Tableau) and modern AI/marketing tools (HubSpot, Canva, NotebookLM, etc.).
• Resume and LinkedIn will be sent.

Target roles (all in scope - apply to any that fit)

1. Sales Development Rep (SDR) / Business Development Rep (BDR)
2. Marketing Coordinator / Marketing Associate
3. Customer Success Associate / Client Success Coordinator
4. Sales Operations / Revenue Operations Analyst entry / Analyst-I level only
5. Junior Salesforce Administrator / CRM Analyst
6. Adjacent early-career roles: growth marketing, demand generation, account coordinator

Location: San Francisco / greater Bay Area. Remote, hybrid, and onsite all acceptable.

Compensation: Minimum $80k

We are building a test set for software that checks schematics for power mistakes. You will get
open-source KiCad 10.0.6 projects. You make copies, introduce one realistic power error into each
copy, and document exactly what you changed and why it is wrong.

Error types (we set a quota per type per batch):
- a supply pin disconnected (including one of several VDD pins, VDDA, VBAT);
- a ground pin or exposed thermal pad disconnected;
- a break in a power path (a ferrite, fuse, 0-ohm link or jumper connection removed);
- a hidden split: a power label misspelled so it looks connected but is not (+3V3 vs +3.3V);
- a wire that ends on a pin without connecting to it;
- a hierarchical/sheet label renamed so a rail stops crossing sheets;
- a part moved onto the wrong rail (a 3.3 V part on 5 V);
- supply and ground swapped, or a supply pin tied to ground;
- two power outputs tied together;
- a series part reversed (a protection diode flipped, regulator input and output swapped);
- a regulator enable pin left floating;
- a feedback divider broken or its values changed so the output voltage is wrong;
- the wrong source part: a regulator swapped for a different-voltage variant, or for a part whose pins are numbered differently;
- a pin the datasheet says must be left open wired to a rail.

We also need clean copies with no error, and decoys: copies with a change that looks
wrong but is not (for example a no-connect pin the datasheet allows to float).

What you deliver per copy:
1. the modified KiCad project, saved in KiCad 10.0.6 only;
2. a PDF export of the schematic with the error circled (mark up the PDF, never the schematic);
3. one row in the error log (template provided): the error type, the parts and pins affected,
the net before and after, the exact edit, why it is wrong with the datasheet page, what would
happen on a real board (no power, damage, wrong voltage, degraded, or nothing for a decoy),
and whether KiCad's Electrical Rules Check flags it.

Rules that matter:
- One error per copy unless the batch asks for more.
- Every error must be realistic: something a tired engineer could actually do.
- Change nothing else: no tidying, re-annotating, moving parts or updating symbols.
- Record ERC honestly. Errors ERC does not catch are the most valuable ones.
- Do not use AI tools to write the explanations.

Acceptance criteria (definition of done):
1. We compare each copy's netlist with the original. The only electrical change must be the one
in your log row; any other change rejects the copy.
2. Each log row names every affected part and pin, the nets before and after, and cites the
datasheet page for "why it is wrong".
3. The batch meets the per-type quota, including clean copies and decoys.
4. Every project opens without errors in KiCad 10.0.6.

We start with a paid trial of 5 copies on one board.

## TTRPG / D&D 5e Adventure Writer Needed for Commercial One-Shot

We are looking for an experienced **TTRPG adventure writer/designer** to create an original, polished fantasy one-shot for commercial publication.

This is not simply a fiction-writing project. We need someone who understands **adventure structure, pacing, encounters, player choice, GM guidance, and how to make a tabletop adventure enjoyable and easy to run**.

### Project

The adventure will be:

* Family-friendly
* Beginner-friendly
* Designed for D&D 5e or a similar fantasy TTRPG structure
* Approximately 2 hours of play
* Intended for commercial publication
* Built around a concept and framework that we will provide privately

We already have the overall product direction and several key elements defined. The selected writer will help turn the concept into a complete, cohesive, playable adventure.

### What We're Looking For

Ideally, you have experience with:

* D&D 5e or similar TTRPG systems
* Writing complete one-shots or adventure modules
* Encounter and adventure design
* Beginner-friendly GM instructions
* Puzzles, exploration, roleplay and combat
* Strong pacing and memorable storytelling

Published work on DMsGuild, DriveThruRPG, Kickstarter, itch.io, or other TTRPG projects is a major plus.

### Requirements

The work must be:

* 100% original
* Human-written — no generative AI writing
* Suitable for commercial publication
* Delivered with appropriate commercial usage rights
* Editable by us for final formatting and product integration

### Timeline

Ideally completed by the **first week of October 2026**.

### When Applying

Please include:

1. A short summary of your TTRPG writing experience.
2. Links or samples of 1–3 adventures/modules you've written.
3. Your estimated delivery timeline.
4. Your fixed-price quote.
5. Confirmation that you're comfortable with commercial publication.
6. Confirmation that you do not use generative AI for the writing.

Please start your proposal with **“Family Adventure”** so we know you've read the full post.

There may be opportunities for additional TTRPG projects if this collaboration goes well.

ACTIVE RECRUITING - PERFECT TO BE YOUR FIRST PROJECT

*Please do not apply if you have already participated in another project from the same organization involving the same or similar activities. Only one active contract per participant is allowed. Previous participation may be verified* AND *You must be in Latin America*

We are looking for reliable contributors to record natural, everyday activities for an ongoing AI data collection project. No previous experience is required.

What you’ll do:
• Record everyday activities.
• Complete a minimum of 8 hours per week (15 hours is ideal)
• Follow simple recording guidelines
• Have a reliable internet connection

Compatible phones:
iPhone 12–17 (except 16e & 17e), Google Pixel 6–9, or Samsung Galaxy S21–S26 series.

Referral Program:
Refer someone who joins the project and completes their first 20 valid recording hours, and receive $20 USD for each successful referral.

If you’re interested, apply and we’ll provide the next steps.

We are building a hand-checked dataset of how power reaches every part on real circuit boards.
You will get open-source board schematics as KiCad 10.0.6 projects, with a PDF of each schematic
and a parts list. For each board you record where power comes from, what every pin does with
power, and the path power takes to reach every part.

What you deliver per board (spreadsheet templates provided):
1. Sources - every place power enters or is created: regulator and converter outputs,
battery terminals, connectors where external power comes in (USB, barrel jack, header), test
points or pads used to feed power. For each: its nominal voltage, how that voltage is set (a
fixed part variant, or feedback resistors and their values), and what turns it on.
2. Pins - for every pin of every part: power in, power out, ground, not connected, or not a
power pin. For an unconnected pin, say whether the datasheet allows it to be left open.
3. Paths - for every power-in pin, the chain of parts back to its source, hop by hop (for
example U5 VDD <- FB2 <- U3 VOUT). For every ground pin, the chain back to the ground of the
source that powers it, including any split grounds joined by a ferrite, 0-ohm link or jumper.

Rules that matter:
- Account for power from every source, including connectors and batteries.
- Trace through passive parts: resistors, 0-ohm links, ferrite beads, inductors, fuses, jumpers
and solder jumpers.
- Trace through regulators, diodes and switches using their datasheets. If a path depends on a
control pin or switch position, record that condition.
- A part marked do-not-fit breaks the path on this build: record the path and mark it broken.
- Connector pins and transistor pins are "not a power pin". A connector can still be a source.
- Decide from the datasheet for the exact part, never from names. Net names, symbol pin types
and power-symbol arrows are often wrong or misleading. If a symbol is a module but the
datasheet is for the bare chip (or the reverse), flag it.
- A pin marked "no connect" on the symbol is judged by the datasheet: fine if the datasheet
allows it open, a fault if the datasheet says it must be powered or grounded.
- Do not edit the schematics. Do not use AI tools to fill in answers. Some boards contain
pins with known answers used to score accuracy.

Acceptance criteria (definition of done):
1. Every pin of every fitted part has a row in Pins; no pin is missing or duplicated.
2. Every power-in and ground pin has a complete path ending at a listed source, or is marked
"no source found" with a one-line reason.
3. Every source has a voltage and how it is set, or "not stated" with the reason.
4. Every datasheet-based decision cites the datasheet page.
5. Accuracy of at least 95% on the unmarked known-answer pins.
6. Delivered in the provided templates, one set per board.

We start with a paid trial on one small board. Workers who pass get the larger boards.

I am looking for an experienced no-code or low-code developer to build a polished, subscription-based web application.

The project will involve:

* Secure user accounts and private dashboards
* Private document uploads and structured records
* AI-assisted document extraction
* Stripe subscriptions
* PDF and email generation
* Responsive desktop and mobile design
* Secure database and API integrations

Experience with Bubble, Supabase, OpenAI API and Stripe is preferred. Privacy, visual quality and a simple customer experience are especially important.

The developer will work independently with me as the founder. I will provide the product concept, priorities and visual direction. I need the developer to advise me on the best technical structure and explain decisions clearly without unnecessary technical language.

Please send examples of relevant products you personally built, briefly explain your role in each and identify which platform you would recommend.

Further product details, deliverables and expectations will be shared privately with selected candidates.

We are building a dataset of how electrical power passes through common components, read from
their datasheets. You will get a list of parts (manufacturer part numbers) with a datasheet for
each. For every part, you record each "channel": a pair of pins power can flow between, which
way it flows, and what has to be true for it to flow.

Examples of what we mean:
- A diode: anode to cathode, one way only, always.
- A linear regulator: VIN to VOUT, one way, only when EN is high.
- A load switch: IN to OUT, only when the control pin is in its "on" state.
- A fuse, ferrite bead or 0-ohm link: both ways, always.
- A TVS or ESD clamp: cathode to anode only above its breakdown voltage (record the volts).
- A mechanical switch: between the contacts, only in a stated position.

What you deliver per part (spreadsheet template provided):
- the component class (diode, schottky, TVS, regulator, load switch, MOSFET, fuse, ferrite, connector, switch, relay, IC, module ...);
- one row per channel: pin A, pin B, direction (A to B, or both ways), condition (none / control pin and its active state / switch position / above a voltage / depends on what is fitted), series or not;
- every other pin accounted for: either a control pin, or "no channel";
- for every row, the datasheet page and a short quote or figure reference that shows it.

Rules that matter:
- A diode is never "both ways". Check direction against the package drawing, not only the pin
names; a reversed diode is the most common mistake we see.
- If the datasheet does not let you decide, write "can't tell". Never write "no channel" for a
pin you could not place.
- Only record paths the part is designed to conduct. Ordinary diodes break down at some voltage,
but that is failure, not function: leave it out. Zeners, TVS and ESD parts are the exception.
- Check the datasheet is for the exact part. Modules and bare chips, and different package
variants, often have different pin numbers. If it is the wrong document, flag it and stop.
- Do not use AI tools to fill in answers. Some parts in every batch have known answers and are
used to score accuracy.

Acceptance criteria (definition of done):
1. Every pin of every part appears in exactly one of: a channel, a control pin, "no channel", or "can't tell".
2. Every channel and every "no channel" has a datasheet page and quote/figure reference.
3. Diode-class channels are one-way, with the direction matching the datasheet drawing.
4. Accuracy of at least 95% on the unmarked known-answer parts in the batch.
5. Delivered in the provided template, one file per part, no renamed columns.

We start with a paid trial of 10 parts. Workers who pass get batches of 50.

PROJECT SCOPE
15 total designs across 3 major holiday events:

- 5 x Thanksgiving designs (Nov 26, 2026)
- 5 x Christmas designs (Dec 25, 2026)
- 5 x New Year designs (Dec 31, 2026)

Each event includes:
- 2 x Quote/typography design
Text will be provided by us, your job is the typography layout, style and visual execution only
- 3 x Illustrated/graphic designs
You create the concept and artwork based on the event theme and brand brief provided after NDA

Mix of fun/humorous and premium/elegant styles required.
No clipart or template-based work.

FILE DELIVERABLES — MANDATORY
---------------------------------------

Every design must be delivered in ALL THREE formats:

1. PNG
- Transparent background
- 300 DPI minimum
- 4500 x 5400 pixels minimum
- sRGB colour mode
- Print-ready, no bleed or crop marks

2. SVG
- Fully scalable vector file
- All text converted to outlines
- Clean paths, no unnecessary nodes

3. AI (Adobe Illustrator) OR Canva Source File
- Fully editable source file
- All layers named and organised
- Fonts embedded or converted to outlines
- Text, colours and graphics must remain editable
- Client must be able to swap quote text independently

PNG + SVG are mandatory regardless of software used.
If you cannot deliver all three formats please do not apply.

TECHNICAL REQUIREMENTS
---------------------------------------

- 300 DPI minimum resolution
- Transparent background on all files
- sRGB colour mode for PNG (DTG/POD printing)
- 4500 x 5400 pixels minimum canvas size
- Maximum print area 12 x 16 inches
- Must work on both light and dark coloured garments
- All designs must be print-ready with no additional preparation needed.

A client is looking for building a sales platform and we're looking for several developers. 3 backend and 3 frontends. Stack needed (You don't need to know them all) because some others devs might have them.
- NodeJS, PHP, Laravel, C#
- React, Vue, Angular
The stack is still undecided as it will be selected from the candiates that have more experience.

It's ok to use AI to develop, in fact we would prefer it as this project will flow more smoothly in terms of timing if AI is used.

Please apply if you have had experience creating sale platforms and similar websites. This project involves handling carts, stock, back orders and suppliers, as well as a section for handling bids, so it's important that you can design solutions to handle high traffic.

For better and faster delivery, we will use a tool called www.autodevtool.com that will be used for automating the design, backend and frontend development so the candidates can spend more time planning and refining requierments. so It's important that the candidates have experience with this tool.

When replying include if you have experience using the tool and how much time you've been using this tool, 1, 2 or 3 years.

I need an original, unique logo designed for a new children’s book brand. The main goal is high recall potential. The design needs to be instantly recognizable and memorable for kids, parents, and educators—whether it is printed on a tiny book spine or viewed on a website.Key Requirements:Concept: 100% original design reflecting the brand name. No generic stock shapes.Versatility: Must look crisp in full color, black-and-white, and scaled down to tiny icon sizes.Required Delivery Formats:Vector Source Files: Fully scalable AI and EPS formats (CMYK for print, RGB for web).Raster Files: Transparent PNG and JPEG formats.Resolution: High-resolution print files at 300 DPI minimum.Please reply with a link to your portfolio, your estimated timeline, and your rates.

Skills: Graphic Design, Logo Design, Photoshop, Illustrator, Branding, Illustration, Creative Design, Book Cover Design, Character Design, Art Direction (Bijutsu Kantoku)

I manually made a33-page work-related website in WordPress for engineering services, 90% competed. I need to have it reviewed and tested. Most consultants want to redo from scratch AI-powered of not, charge $250 a month till the end of times, not interested. We would like to continue hosting ourselves. The task would be to review and test the new 33 page website, provide comments so that I can fix it on my time, test it for optimal performance so it is ready to launch.

Talligen Systems, an IT infrastructure consulting business, is looking for a female voice actor to record a short professional voicemail greeting.

The finished recording will be approximately 20 seconds and 40–45 words.

I'm looking for a natural, warm, confident, conversational voice with clear diction and an understated professional delivery. It should sound like a real person representing a professional consulting firm—not a radio commercial, corporate announcer, or overly enthusiastic customer-service recording.

I'm particularly interested in British or Australian English voices, although voice and delivery are more important than a specific accent.

This must be a human voice recording. Please do not submit AI-generated or text-to-speech audio.

The business name is Talligen Systems. Talligen is pronounced with a short “a” as in pal: TAL-ih-gen.

Deliverable is one clean, edited WAV recording with no music or sound effects, suitable for a business telephone/voicemail system. Please include one reasonable pickup/revision if needed for pronunciation, pacing, or delivery, along with commercial/business usage rights for the finished recording.

Please include a voice demo or sample of previous work with your proposal. A sample demonstrating natural business/telephone delivery would be especially helpful.

Script

Thank you for calling Talligen Systems. No one is available to take your call at the moment. Please leave your name, telephone number, and a brief message. We will return your call as soon as possible. Thank you.

We are looking for an Azure certified instructor to record training videos. All content and lab material will be provided. The certified trainer will be required to record training videos, where all the training content, and the content will be provided any errors in the lab content and lessons will have to be fixed during the course of recording mostly it is expected that the content is fine as provided please review the topics below and only expert AZURE trainers are required to reply to this job.


Please review the outline of the course listed below

THIS CONTENT MAPS TO THE AZ 900 AND AI 200 CERTIFICATIONS FOR AZURE.

*********

===============================================================================
AZURE FOUNDATIONS AND AI CLOUD DEVELOPMENT
Course Overview and Lesson List
================================================================================

ABOUT THIS COURSE
--------------------------------------------------------------------------------
Azure Foundations and AI Cloud Development is a single twenty-lesson course that
takes a complete beginner from cloud concepts to a running, containerised,
event-driven AI back end on Azure.

It harmonises two certification blueprints into one dependent arc rather than
teaching them as two half-courses. The foundational platform layer — cloud
concepts, architecture, storage, identity, governance, and monitoring — is taught
first and then used, not restated, as the build layer for AI cloud development:
container registries and hosting, serverless functions, messaging and eventing,
NoSQL and relational vector data services, caching, and retrieval-augmented
generation. Where the two blueprints touch the same surface, the topic is taught
once and deeply at the earliest point it is needed.

No prior cloud, container, or Azure experience is assumed. Every tool used is
free and runs on Windows, macOS, and Linux, and every lab is sized for free-tier
and free-credit allowances.

Each lesson ships a twenty-slide deck, a hands-on lab guide, a verbatim
transcript, an instructor guide, a lab answer guide, and a lab assets package
containing the scripts, templates, and starter files that lab requires.

THE CAPSTONE — "SEMANTIC HELPDESK"
--------------------------------------------------------------------------------
A support organisation is drowning in repeat questions whose answers already
exist, buried in years of accumulated support articles. Across twenty labs you
build the system that finds them: documents land in storage, an event queues each
one for processing, a serverless function extracts and stores it with its
embedding vector, and a containerised API answers questions by retrieving the
most semantically similar passages and grounding its response in them. Identity,
secrets, cost controls, governance, scaling rules, and distributed tracing are
added as the platform layer is taught. Every lab consumes the artifact the
previous lab saved.

--------------------------------------------------------------------------------
THE TWENTY LESSONS
--------------------------------------------------------------------------------

LESSON 1 — Cloud Computing, Cloud Models, and Shared Responsibility
What the cloud delivers, who answers for each layer of it, and which
responsibilities never transfer to a provider.

LESSON 2 — Working in Azure: Portal, Cloud Shell, CLI, and PowerShell
Four ways into the platform, when each is the right one, and the
local toolchain every later lab depends on.

LESSON 3 — Cloud Benefits and Cloud Service Types
Turning advertised benefits into usable criteria, and mapping the
three service types onto the responsibility boundary.

LESSON 4 — Azure Global Infrastructure and the Resource Hierarchy
Where a resource lives and what it belongs to: regions, zones,
datacenters, and the four-level scope hierarchy.

LESSON 5 — Azure Compute and Application Hosting Options
Choosing between machines, containers, and functions, and deciding
how each capstone component will be hosted.

LESSON 6 — Azure Networking, Connectivity, and Endpoints
Virtual networks, subnets, peering, name resolution, hybrid
connectivity, and what a private endpoint changes.

LESSON 7 — Azure Storage Services, Redundancy, and Data Movement
Comparing storage services, choosing tiers and redundancy, and moving
the document corpus into the cloud.

LESSON 8 — Identity, Access, and Security Posture
Directory services, authentication methods, conditional access,
role-based access control, and defence in depth.

LESSON 9 — Cost Management, Governance, and Resource Deployment
What drives a bill, how to constrain an estate with tags, policy and
locks, and deploying resources declaratively from a template.

LESSON 10 — Monitoring, Log Analysis, and Distributed Tracing
Collecting telemetry, writing queries against it, and following a
single request across components by trace identifier.

LESSON 11 — Container Images and Azure Container Registry
Building images, versioning them with tags and digests, and running
registry-side builds.

LESSON 12 — Hosting Containers on App Service and Azure Container Apps
Running a container image on two managed platforms, supplying its
secrets safely, and splitting traffic across revisions.

LESSON 13 — Serverless APIs with Azure Functions
Triggers, bindings, and configuration for code that runs on an event
and costs nothing between executions.

LESSON 14 — Event- and Message-Based Workflows with Service Bus and Event Grid
Queues, topics, dead-letter handling, and event-driven workflows that
connect the pieces of the system together.

LESSON 15 — Scaling and Operating Containers with KEDA and Kubernetes
Scaling on queue depth, deploying to Kubernetes from manifests, and
diagnosing a broken containerised workload.

LESSON 16 — Azure Cosmos DB for NoSQL: SDK, Queries, Indexing, and Change Feed
Connecting with the SDK, controlling query cost through indexing and
consistency, and reacting to changes as they happen.

LESSON 17 — Vector Search and Caching with Cosmos DB and Azure Managed Redis
Storing embeddings, running similarity search, and caching results
with sensible expiry and invalidation.

LESSON 18 — Azure Database for PostgreSQL for Vector Workloads
Schema and index design for vector data, sizing the server for it,
and optimising connections for throughput.

LESSON 19 — Retrieval-Augmented Generation with pgvector
Semantic retrieval with metadata filtering, and grounding answers in
the passages the system actually found.

LESSON 20 — Capstone Completion and What's Next
Assembling the finished Semantic Helpdesk end to end, documenting it,
verifying operational readiness, and where these skills lead.

--------------------------------------------------------------------------------
HOW THE COURSE PROGRESSES
--------------------------------------------------------------------------------
Lessons 1-4 Concepts and first resources — vocabulary, tooling, and the
resource group everything else is built inside.
Lessons 5-8 The platform layer — compute, network, storage, and identity.
Lessons 9-10 Controlling and observing the estate you have built.
Lessons 11-15 Packaging, hosting, connecting, and scaling the application.
Lessons 16-19 The data and retrieval layer that makes the system intelligent.
Lesson 20 Assembly, documentation, and handover.

================================================================================
END OF COURSE OVERVIEW
================================

We are looking for confident, natural on-camera creators to produce promotional videos showing their experience using en2uition, an AI-guided self-reflection app designed to help people understand their romantic relationships more clearly.

## What You Will Do

Each creator will be assigned:

* A specific level of the app from among six available levels
* A specific topic within that level
* Either a past-relationship or current-relationship perspective
* Instructions about the main parts of the experience to feature

Each topic contains approximately 15 questions. You will first spend time using the assigned part of the app so that you understand the questions, the experience, and the feedback you receive.

Your recorded video should include a brief introduction explaining what the assigned topic explores. You may then show yourself reading selected questions, answering some of them aloud, choosing answers, skipping a question, or returning to a previous question. The interaction should feel natural and reflect how someone would realistically use the app.

You do not need to show every question or reveal personal information that you are uncomfortable sharing.

## Exploring the Feedback

After answering the questions, the app provides personalized feedback across three tabs. Each tab contains four sections.

Your video may show the feedback appearing on the screen. You can then pause the recording, take time to read it carefully, and make notes about:

* What impressed or surprised you
* What felt accurate
* What was personally useful
* Which observations made you think differently
* Which reflective questions stood out
* Where you wanted more detail
* Which sections felt more or less relevant to you

When recording your reaction, you may read selected parts of the feedback or some of its reflective questions aloud. We want a thoughtful personal response—not simply a demonstration of the screen.

## Honest and Balanced Reactions

We do not expect every reaction to be completely positive. Thoughtful uncertainty or mild criticism can make the video more credible, but it must reflect your genuine experience.

If it matches your actual reaction, you might explain that:

* A particular observation could have explored the issue more deeply.
* You would like the option to ask follow-up questions.
* Some of the feedback was longer than you personally needed.
* You preferred one section and chose to move quickly through another.
* One observation did not fully fit your situation, while another felt highly accurate.

You can balance an honest criticism by explaining who might benefit from the feature or what remained useful. However, you should never make a statement you do not genuinely believe. We want natural, credible, and balanced content—not exaggerated praise or artificial criticism.

## Preparation and Recording Time

Before recording, you should spend approximately 30–45 minutes using the assigned part of the app and reading the feedback. The recorded video should not capture the very first time you see the app or its feedback.

After becoming familiar with the experience, you may recreate selected moments for the camera. This allows you to speak clearly, choose the most useful sections, and produce a focused video while keeping your reactions truthful.

You should expect to spend another 45–90 minutes planning and creating the video. The total expected time for each assignment is approximately 1.5–3 hours.

Video Length and Format

The finished video should be between 2 and 3 minutes long, with an ideal duration of approximately 2 minutes and 30 seconds.

The video should be recorded vertically in a 9:16 format and delivered in high resolution, preferably 1080 × 1920. It should combine natural on-camera commentary with clear footage or screen recordings of the app.

You do not need to show every questionnaire item or read every feedback section. Briefly show the overall experience, then focus on the questions, observations, and feedback that produced your most interesting or meaningful reactions.

The video should feel thoughtful and complete without becoming repetitive. We may later create shorter edits from the finished video for TikTok, Instagram, Facebook, and paid advertising.

## What We Are Looking For

The ideal creator:

* Is comfortable speaking naturally on camera
* Can record clear video and audio
* Can communicate emotional or personal ideas thoughtfully
* Can follow assignment-specific instructions
* Can give an honest and balanced opinion
* Can make an app demonstration feel engaging and relatable
* Has experience with UGC, TikTok, Instagram Reels, testimonial-style videos, or promotional content

Individuals, experienced UGC creators, and agencies may apply. Agencies may propose multiple creators, but each person who appears in a video must be presented to us and approved before receiving an assignment.

## How to Apply

Please include:

1. A short introduction about yourself
2. Links to two or three relevant videos
3. Your experience creating UGC or speaking on camera
4. Whether you are applying as an individual creator or an agency
5. Whether you can produce multiple videos
6. How many videos you could realistically complete
7. The recording equipment and editing software you use
8. Your expected turnaround time for each video

Please begin your proposal with the word “en2uition” so we know you read the complete description.

Each assignment will include a specific level, topic, relationship perspective, and creative brief. Payment will be released after the submitted video follows the brief and is approved.


Payment and Ongoing Work

Payment is $250 for each completed and approved video.

An additional performance bonus of up to $100 may be awarded for an exceptional video. The bonus may reflect outstanding authenticity, presentation, creativity, recording quality, editing, emotional insight, and overall usefulness for our campaign. The bonus is discretionary and is not guaranteed.

We are hiring multiple creators, and successful creators may be offered five or more video assignments. Each assignment will cover a different level or topic within the app and will be funded separately.

The $250 payment includes the preparation time, recording, editing, captions, delivery of a clean watermark-free video, and one reasonable revision based on the original creative brief. Additional deliverables, creator-account publication, paid-ad authorization, or extended advertising rights will be negotiated separately.

##### if you want to explore our app before applying here it is:
www.en2uition.com

https://en2uition.com/how-it-works

Category / Skills

Graphic Design, Print Design, Marketing Collateral, Brand Identity, Adobe Illustrator, Adobe InDesign, Canva

Job Description

We need a modern, eye-catching postcard/handout designed for our booth at Dreamforce (Sept 15–17, Moscone Center, San Francisco) — and we need it fast. This is a rush project: work happens over this weekend, with final files delivered Sunday night.

What we need: A single-piece handout postcard front and back that we can print and hand out at our booth. It needs to grab attention on a busy expo floor, communicate our value prop at a glance, and drive people to scan a QR code or visit our website.

Design direction:

Match the look and feel of our website. Modern, clean, and attention-grabbing. NOT busy or cluttered. Polished, not gimmicky. We'll provide our logo, brand assets, and key messaging/copy points; you bring the layout, typography, and visual execution (a strong minimalist graphic or icon treatment is a plus)

Deliverables:

Print-ready PDF (with bleed/crop marks) at [postcard: 4x6in / handout: 5x7in or 8.5x11in - double-sided
A web/digital version (PNG or JPG) for email/social use
Editable source file (AI, INDD, or Figma/Canva link)

Timeline (firm):

Awarded: today
Draft concept: Saturday
Revisions: Saturday evening–Sunday
Final files: Sunday night, no later than 9pm PT

Please only apply if you're realistically available to turn this around this weekend.

To apply, please include:

2–3 examples of past print/trade-show marketing design work
Confirmation you can deliver final files by Sunday night

We're looking for salespeople to spend two weeks stress-testing AI models against the work you actually do. You'll write realistic task scenarios, run them at the model, and build the rubrics other reviewers grade against.

About the role
This is a two-week pilot campaign built around AI model evaluation, task design and grading frameworks.

The point is to benchmark next-generation models by giving them complex, realistic, domain-specific work. You'll design tasks that need multi-step research, data extraction and real professional judgment, then set the criteria for scoring them.

Full-time for the duration of the contract.

What you'll do
Complex prompt creation
Write detailed scenario prompts rooted in real sales workflows: account research and prospect qualification, pipeline review and forecasting, quote and proposal construction, CRM hygiene and deal-stage auditing, territory planning, and post-call follow-up sequencing. Some briefs run through QuickBooks, covering estimates, invoicing and customer records.

AI model testing
Run your prompts in the platform to surface failure modes, hallucination triggers and reasoning gaps. Escalate the complexity as you go.

Rubric and checklist authoring
Write objective, self-contained scoring checklists backed by explicit reference sources, so peer reviewers grade the same output the same way.

Requirements
Domain expertise
5+ years in a sales role: AE, SDR leadership, sales ops or revenue management.

AI training experience
Hands-on experience writing complex prompts, evaluating LLM output and building evaluation rubrics.

Language
Professional English, written and spoken.

Availability
Full-time across the full two weeks.

Device security
Work happens on a secure personal or corporate device. Public or shared computers aren't permitted.

Nice to have
Not required. Worth mentioning if it applies to you.

You've carried a quota and also run forecasting or pipeline review.
Experience in more than one CRM, or migrating between them.
Hands-on QuickBooks experience: estimates, invoicing or customer records.
Enterprise or complex multi-stakeholder deal experience.
You've written playbooks, call frameworks or onboarding material for other reps.

Why take it
Your judgment sets the bar. The rubrics you write are what everyone else grades against.
You'll see where frontier models actually break on work you know cold.
Two weeks, remote, with the fee agreed before you start.

Contract and payment
You're engaged as an independent contractor for the length of the campaign.
Fully remote, from any location, working in English.
The fee is agreed before you start. We pay you directly, not through an agency.
You won't use your employer's or your clients' data. The tasks are scenarios you author. Don't bring confidential information to this work.
Campaigns can be extended, shortened or ended early depending on need and performance.

About Askable
Askable is research infrastructure, founded in Brisbane in 2017. Askable Labs, a division of Askable Pty Limited, produces primary data for frontier AI labs from experts doing real work. We recruit those experts ourselves, verify them, and pay them directly.

We consider every qualified applicant without regard to legally protected characteristics, and we make reasonable adjustments on request.

Skills: Sales, CRM, Intuit QuickBooks, Sales Management, Data Analysis, Prompt Engineering, AI Model Development, AI Training Data, AI Workflow Automation

I need a concise, unilateral non-disclosure agreement that shields all intellectual property connected to my brooch design work—from Ai and CAD files to production techniques and supplier details. The document should be plug-and-play: I want to be able to drop in the counter-party’s name, the effective date, and sign.

Please make sure it:

• Defines “Confidential Information” broadly enough to cover current and future design iterations.
• Sets out clear obligations on the receiving party and a reasonable but firm duration of confidentiality.
• Includes standard exclusions (public domain, prior knowledge, compelled disclosure) and an injunctive-relief clause.
• Leaves governing law and venue as easy-to-edit placeholders.
Company is a Chinese Manufacturer
Deliver the final draft in an editable Word file (plus a clean PDF for immediate use).

Skills: Editing, Legal Research, Drafting, Word Processing, Business Writing, Legal Writing, Business Consulting, Contract Management, Legal Analysis, Intellectual Property Law

We’re looking for a graphic designer / motion designer to create the official cover artwork and a simple animated promotional version for the first release on Hale Records, an independent underground house/techno label.

Project:
Artist: Russell Trent
Track: APU
Label: Hale Records
Catalog: HR001

The static cover artwork is the primary deliverable and will be used for digital distribution through platforms such as Beatport, Spotify, Apple Music, SoundCloud, and other DSPs. We also want a simple, tasteful animated version of the approved artwork for social media promotion.

Visual direction:
- Sophisticated
- Underground
- Modern
- Atmospheric
- Minimal
- Slightly futuristic
- Possible inspiration from machinery, activation, energy, turbine geometry, power systems, or systems coming online

We want to avoid generic EDM/festival artwork or overly commercial visual treatment.

We already have a finalized professional Hale Records logo/brand system that will be provided to the selected designer. Russell Trent also has an existing artist logo if needed.

Expected deliverables:

Static artwork
- 3000 x 3000 px square cover
- RGB JPG suitable for digital distribution
- High-resolution PNG
- Editable/source file (PSD, AI, or equivalent)
- Commercial-use rights

Animation
- Simple animation derived from the approved cover artwork
- Approximately 10–15 seconds
- 1080 x 1920 vertical MP4 for Reels/Stories
- 1080 x 1080 square MP4 for social media/feed use
- Editable/source motion file where applicable
- Animation should be subtle and does not need to be a complex cinematic production

Ideal candidate:
- Strong portfolio in electronic music artwork, preferably underground house, techno, melodic techno, deep house, or related genres
- Experience with both graphic design and basic motion graphics
- Strong visual composition and typography
- Able to work within an established brand system
- Responsive to feedback without unnecessarily redesigning approved elements
- Clear communication and reliable delivery

Budget:
Up to $200 fixed price total for the complete project.

This may lead to additional Hale Records release artwork and promotional work if the collaboration is successful.

Please include:
1. Relevant cover-art examples
2. Relevant animation/motion examples
3. Confirmation that you can provide editable/source files
4. Your proposed timeline for completing both deliverables

Take ~100 photos in Google Photo album to create a slide show. Music will be provided as well. About 30 photos need minor touch-ups or rotations. Some need light AI restoration because they were taken of old photos. The rest of the photos are digital photos that can be used as is. They are grouped into 2 albums: need touch up and ready to use.

We are expanding our data collection operations in Argentina and are seeking detail-oriented individuals to participate in a remote video research study. The objective of this project is to build diverse visual datasets that will train next-generation Artificial Intelligence (AI) models and machine learning systems to better understand human movement and daily physical tasks.


🔄 Your Role:

You will use your smartphone attached to a head strap to film short, first-person perspective (POV) videos of your regular domestic routines. This ensures you capture the tasks exactly from your point of view while keeping your hands entirely free.

🔒 Absolute Privacy Guaranteed:

Your personal identity is fully protected. All submitted footage focuses strictly on your hands and the specific manual tasks you perform. Your face will never be captured on camera, and audio recording is completely unnecessary.

💰 Earnings & Schedule:

- Compensation Rate: US$4 for every completed hour of fully approved footage.
- Contribution Limit: Each participant can submit up to 100 hours of video.
- Maximum Earnings: Up to US$400 total payout per contract.
- Schedule Flexibility: There are no rigid shifts or fixed deadlines. You are completely in charge of your own time and can record whenever it fits your personal routine.

📱 Daily Activities You Will Film:

You will record natural, unscripted domestic habits (usually lasting between 10 to 30 minutes per clip). Typical tasks include:

- Making breakfast, lunch, dinner, or preparing quick snacks.
- Brewing a cup of coffee, mate, or tea, as well as baking at home.
- Kitchen maintenance, washing dishes, or wiping down countertops.
- General home care, sorting household items, folding laundry, or unpacking groceries.
- Everyday pet care, such as feeding or grooming your pet.

⚠️ MANDATORY HARDWARE ELIGIBILITY:

To guarantee technical compatibility with our AI models, your smartphone MUST feature a native ultra-wide (0.5x) camera lens. We strictly accept only the following devices:

- Apple: iPhone 12 through iPhone 17 lines (Includes Mini, Plus, Pro, and Pro Max models. Note: 16e and 17e models are strictly excluded).
- Samsung: Galaxy S21 through Galaxy S26 series (Includes Plus and Ultra editions).
- Google: Pixel 6 through Pixel 9 lineups (Includes Pro models).

* Technical Note: A stable Wi-Fi connection and a secure head strap/mount are mandatory requirements for participation.

🇦🇷 Participant Eligibility:

- You must physically and legally reside full-time within Argentina.
- You must maintain high precision when following our required framing, visual quality guidelines, and specific camera angles.
- All actions on camera must display genuine, authentic, and non-staged behavior.

🚀 How to Apply:

Please keep your Upwork proposal brief and answer these 3 questions directly:
1. What city and province in Argentina do you currently live in?
2. What is the exact manufacturer, model, and series of your smartphone?
3. How many hours per week can you realistically dedicate to this project?

⏳ REQUIRED PROJECT COMMITMENT:

To keep your contract active and in good standing, you must deliver a minimum of 2 hours of approved video data daily or at least 15 hours every single week. Failing to hit this weekly baseline volume will put your contract at immediate risk of suspension or cancellation. Please do not apply if you cannot guarantee this specific workload.

Build Something Meaningful With Us

We’re not hiring just another assistant. We’re looking for someone who wants to grow with a company, contribute ideas, take ownership, and be part of a team building something exciting from the ground up.

We are creating a community centered on meaningful connections, wellness, content creation, entrepreneurship, and unforgettable experiences. If you love social media, enjoy helping people, and thrive in a creative, remote environment, we’d love to meet you.

Compensation & Growth
Months 1-4: $600
Months 5-8: $650
Months 9-12: $700

In addition to salary increases, we also offer:

- Performance bonuses
- Additional responsibilities with higher pay
- Leadership opportunities
- Long-term career growth as our company expands

What You’ll Be Doing
Handle phone calls warmly for our beauty brand
Create engaging Instagram Reels, Facebook Reels, long form videos
Help promote brand events and experiences
Respond to client messages and inquiries with warmth and professionalism
Assist with daily administrative tasks
Keep the online community active and engaging
Update the team on progress and key tasks

Who You Are
Confident to handle phone calls with American clients warmly and kindly
Experienced with social media platforms and content creation
Comfortable managing client communications and administrative tasks
Highly organized, dependable, and proactive
Quick learner and eager to take initiative
Clear and professional communicator
Able to work independently while keeping the team informed
Looking for a long-term role with meaningful growth opportunities
Comfortable with Hubstaff, ChatGPT, Canva, CapCut
Has Apple Device for File Sharing
Comfortable to work San Diego Business Hours, 40 hours a week including weekends
Has Flexible schedule

How to Apply
Send an introduction video (1–2 minutes) to ---------- along with:

Your name and location
Social media experience (Instagram, Facebook, TikTok, YouTube, Canva, CapCut, etc.)
Examples of your work (Reels, graphics, websites, landing pages, or presentations)
Experience in handling phone calls

In your video, please tell us:

What excites you about working with our brand
Why you’d be a great fit
Which social media platforms you’re experienced with
A little about yourself

We’re looking for energetic, creative, and professional individuals with a genuine desire to contribute to building something meaningful.

## Job Title

**High-End Direct Response Video Editor Needed for 4 Paid Social Ads**

## Job Description

We’re **12Media**, a marketing agency, and we’re looking for a high-level video editor to create **4 premium short-form paid ads** for two different campaigns:

**2 ads for law firms in Fresno/Central Valley, CA**
**2 ads for peptide/research brands**

We already have the **scripts, branding, and a reference video** showing the exact pacing and quality we’re looking for.

These are NOT basic talking-head edits.

We need someone who understands **direct-response advertising** and knows how to keep viewers engaged from the first second until the CTA.

The editing style needs to be:

* Extremely fast-paced and engaging
* Premium and polished
* Strong pattern interrupts
* Motion graphics and animated typography
* Sound design synced perfectly with the edit
* Whooshes, clicks, static, typing, notification sounds, etc.
* Fast cuts and punch-ins
* Website/interface animations
* B-roll that actually supports what is being said
* Clean subtitles/captions
* Strong hook in the first 1–2 seconds
* Visual changes every few seconds to maintain retention
* Professional color grading
* Strong CTA ending
* Built specifically for paid Meta/Instagram advertising

**We do NOT want AI slop.**

No generic CapCut template, random stock footage, excessive emojis, cheap transitions, or an edit that looks like someone simply added captions to a voiceover.

The final ads need to feel like they came from a **high-end performance creative agency.**

## THE 4 ADS

**LAW FIRM AD #1**
Three-option style ad targeting law firms in Fresno and the Central Valley.

**LAW FIRM AD #2**
Fast-paced direct-response ad focused on law firms spending money on marketing but failing to stand out.

This campaign will feature 12Media's services, including professional video production, websites, advertising, creative, and lead funnels.

**PEPTIDE AD #1**
Targets people launching or operating peptide brands.

Focuses on building the entire infrastructure: website, creative, advertising, SEO, and access to payment-processing options suitable for qualifying businesses in the industry, including businesses starting at $0 revenue.

**PEPTIDE AD #2**
Three-option format highlighting the difficulty of piecing together a peptide business yourself versus working with one team.

## REFERENCE STYLE

We will provide the reference video.

**FOLLOW THE REFERENCE CLOSELY.**

Pay attention to:

Pacing → sound design → typography → transitions → pattern interrupts → B-roll → visual storytelling → CTA.

We don't want someone to completely reinterpret it.

We want the same **energy and level of execution**, adapted to the 12Media brand and our scripts.

## DELIVERABLES

4 finished ads total.

Primary format: **9:16 vertical (1080×1920)**

Each video should be optimized for Instagram/Facebook Reels and paid social.

We need:

* Final MP4
* High-quality export
* Clean version if requested
* Project/source files
* All assets organized
* Reasonable revisions included

## WHO WE'RE LOOKING FOR

You should have experience editing **high-performing paid social ads**, preferably for agencies, professional services, e-commerce, or other direct-response campaigns.

You need to understand the difference between:

**A video that looks cool**

and

**a video designed to stop the scroll and generate action.**

When applying, please send **2–3 of your BEST direct-response ads.**

Do not send us a giant portfolio.

Send the work most similar to the reference style.

Also tell us:

1. Your price for all 4 ads
2. Your turnaround time
3. How many revisions are included
4. Whether source/project files are included
5. What software you edit in

**If your portfolio is mostly basic talking-head captions or generic short-form edits, this probably isn't the right project.**

We're looking for someone we can continue working with on **ongoing paid ad creative for 12Media and our clients**.

**Potential for long-term work if these first four ads are strong.**

Scripts
https://drive.google.com/drive/folders/112Gxucl-rDdbYJ0OpATYhASnU8RTJIrD?usp=sharing

Overview

I run a documentary-style YouTube channel called Uncovered Cases, covering real, verified events crime, disasters, corporate scandals, political corruption, and unsolved mysteries. I need a freelance editor to take a finished script and turn it into a fully edited, publish-ready video.

What I'll Provide
A complete, timed script (cold open, chapters, ending)
Channel logo and branding assets (colors: black, deep red, navy)
A reference video or two showing the visual style I'm going for (similar to channels like Fern)
What I Need From You
Voiceover: Either you provide AI or human narration matching a confident, cinematic documentary tone, or I'll send you a separate voiceover track to sync to — let me know your preference/capability.
Visuals: Stock footage, archival photos/clips, motion graphics, maps, and simple animations to illustrate the script — no talking-head/on-camera footage needed (faceless channel).
Editing: Pacing that keeps retention high — dynamic but not chaotic, matching the tone of the script (serious, investigative, not comedic).
Sound design: Background music (royalty-free/licensed), sound effects, and audio mixing/leveling.
Text/graphics: On-screen captions or key text call-outs where useful, lower thirds for names/dates/locations.
Color grading: Consistent, cinematic grade matching a moody, high-contrast red/black/navy palette.
Final export: 1080p or 4K, YouTube-ready format (MP4, H.264).
Scope & Deliverables
Runtime: 10–30 minutes per video (varies by script)
Delivery of a rough cut for review before final polish
One round of revisions included
Turnaround time: As soon as possible while not letting it take a toll on the quality.
Budget:

$50 dollar per video (budget may rise up to $100 if the videos perform well)

To Apply

Please share:

Your portfolio/reel — ideally similar faceless documentary/YouTube work
Your typical turnaround time per 15-20 minute video
Whether you can also handle voiceover, or if I need to source that separately
Your rate (flat per video preferred)

We're hiring a Direct-Response Ecom Video Editor who knows how to turn raw UGC, winning hooks, and product footage into high-converting ad creative for European markets. If you can take viral trends, supplier clips, and AI assets to build scroll-stopping shorts and full-length ad funnels that scale across the EU, keep reading.

What we're looking for:
-You've edited high-performing direct-response ads specifically for ecom and dropshipping products.
-You understand winning ecom frameworks — pattern interrupts, problem-agitate-solution, offer positioning, urgency, and social proof tailored to EU buyers.
-You know how to assemble high-converting creatives from raw supplier videos, UGC, and competitor mashup content across both short-form (Reels/TikTok/Shorts) and long-form (VSLs/YouTube) formats.
-You can quickly source or generate missing assets, including localized AI B-roll, realistic multi-language AI voiceovers, and dynamic product graphics.
-You're fluent with ElevenLabs and can adjust tone, pacing, accents, and audio design to match diverse European demographics and languages.
-You know how to leverage AI tools for 3D product animations and custom B-roll generation.
-You're fast: Able to deliver rapid turnarounds on new ad concepts, batch hook iterations, and format cutdowns built for European campaigns.

Pay & Structure:
-Project-based to start (get paid per asset/deliverable during the initial testing period).
-Fast-track to full-time: Once your creatives scale campaigns and drive revenue for the business, we move you straight to a fixed monthly retainer + performance bonuses for winning ads.

How to apply (Read carefully):
1- Submit your portfolio with your best ecom and dropshipping video ads (short-form and long-form)
Apply now if you're ready to pump out winning ad creatives with a performance-driven team.

Let's scale.
Best,
Ismail
Founder – Scalix

We are looking to hire an experienced Full-Stack Developer / AI Developer to build an AI-powered SaaS platform for property rental services.

The core product will be an intelligent AI chatbot that helps property rental businesses automate customer communication, answer property-related questions, qualify potential tenants, and assist users throughout the rental process.

The goal is to build a scalable SaaS product that can eventually be offered to multiple property managers, landlords, and rental agencies as a subscription-based service.

What We Expect

Please include the following in your proposal:

Examples of similar AI/SaaS projects you have built
Your recommended technology stack
How you would architect the AI chatbot
Your approach to RAG/property data
Estimated timeline
Estimated project cost or milestone-based pricing
Any recommendations or concerns you have about the proposed product.

We're hiring a Direct-Response Ecom Video Editor who knows how to turn raw UGC, winning hooks, and product footage into high-converting ad creative for European markets. If you can take viral trends, supplier clips, and AI assets to build scroll-stopping shorts and full-length ad funnels that scale across the EU, keep reading.

What we're looking for:
-You've edited high-performing direct-response ads specifically for ecom and dropshipping products targeting European audiences.
-You understand winning ecom frameworks — pattern interrupts, problem-agitate-solution, offer positioning, urgency, and social proof tailored to EU buyers.
-You know how to assemble high-converting creatives from raw supplier videos, UGC, and competitor mashup content across both short-form (Reels/TikTok/Shorts) and long-form (VSLs/YouTube) formats.
-You can quickly source or generate missing assets, including localized AI B-roll, realistic multi-language AI voiceovers, and dynamic product graphics.
-You're fluent with ElevenLabs and can adjust tone, pacing, accents, and audio design to match diverse European demographics and languages.
-You know how to leverage AI tools for 3D product animations and custom B-roll generation.
-You're fast: Able to deliver rapid turnarounds on new ad concepts, batch hook iterations, and format cutdowns built for European campaigns.

Pay Structure:
-$12 per 1-minute edited video
-$2 per hook
-Performance bonuses for winning scaling creatives

How to apply (Read carefully):
1- Submit your portfolio with your best ecom and dropshipping video ads (short-form and long-form) produced for EU markets.
Apply now if you're ready to pump out winning ad creatives with a performance-driven team.

Let's scale.
Best,
Ismail
Founder – Scalix

Looking for English speakers for AI/TTS voice recording project.

You have to record 3-4 minute short recording in different scenarios and we need professional recording equipment for this project.

If you are not interested in AI/TTS related project then don't apply on this job.

About the role
This is a two-week pilot campaign built around AI model evaluation, task design and grading frameworks.

The point is to benchmark next-generation models by giving them complex, realistic, domain-specific work. You'll design tasks that need multi-step research, data extraction and real professional judgment, then set the criteria for scoring them.

Full-time for the duration of the contract.

What you'll do
Complex prompt creation
Write detailed scenario prompts rooted in real sales workflows: account research and prospect qualification, pipeline review and forecasting, quote and proposal construction, CRM hygiene and deal-stage auditing, territory planning, and post-call follow-up sequencing. Some briefs run through QuickBooks, covering estimates, invoicing and customer records.

AI model testing
Run your prompts in the platform to surface failure modes, hallucination triggers and reasoning gaps. Escalate the complexity as you go.

Rubric and checklist authoring
Write objective, self-contained scoring checklists backed by explicit reference sources, so peer reviewers grade the same output the same way.

Requirements
Domain expertise
5+ years in a sales role: AE, SDR leadership, sales ops or revenue management.

AI training experience
Hands-on experience writing complex prompts, evaluating LLM output and building evaluation rubrics.

Language
Professional English, written and spoken.

Availability
Full-time across the full two weeks.

Device security
Work happens on a secure personal or corporate device. Public or shared computers aren't permitted.

Nice to have
Not required. Worth mentioning if it applies to you.
- You've carried a quota and also run forecasting or pipeline review.
- Experience in more than one CRM, or migrating between them.
- Hands-on QuickBooks experience: estimates, invoicing or customer records.
- Enterprise or complex multi-stakeholder deal experience.
- You've written playbooks, call frameworks or onboarding material for other reps.

Why take it
- Your judgment sets the bar. The rubrics you write are what everyone else grades against.
- You'll see where frontier models actually break on work you know cold.
- Two weeks, remote, with the fee agreed before you start.

Contract and payment
- You're engaged as an independent contractor for the length of the campaign.
- Fully remote, from any location, working in English.
- The fee is agreed before you start. We pay you directly, not through an agency.
- You won't use your employer's or your clients' data. The tasks are scenarios you author. Don't bring confidential information to this work.
- Campaigns can be extended, shortened or ended early depending on need and performance.

About Askable
Askable is research infrastructure, founded in Brisbane in 2017. Askable Labs, a division of Askable Pty Limited, produces primary data for frontier AI labs from experts doing real work. We recruit those experts ourselves, verify them, and pay them directly.

We consider every qualified applicant without regard to legally protected characteristics, and we make reasonable adjustments on request.

I’m looking for a freelancer to create a couple of 15-second reels based on existing static ads in my Facebook campaign. The work involves turning static visuals into short, engaging video content that aligns with the campaign’s messaging and style. This is initially a very small job - maybe only 1-2 hours - for someone who is adept at utilising AI and/or traditional graphic design / video editing tools to deliver polished, platform-ready reels to Meta specifications. Hoping to retain the right person for ongoing support with content creation.