Freelance python jobs for world-class Developers

Python + Playwright Scraping Engineer (Argaam KSA + UAE, Multi-Ticker Harvester)
Project summary

I need an experienced Python + Playwright scraping/automation professional to stabilize, refactor, and complete my existing Argaam scraper so it reliably collects a defined set of pages, tables, articles, and PDFs for each ticker.

I already have a draft script (Playwright + persistent profile + manual login/2FA approach). It is not reliable yet. Your job is to make it production-grade: robust navigation, correct selectors, pagination, date-range extraction, consistent outputs, and strong logging/error handling.

Target markets / domains

The scraper must support both:

Saudi market (KSA): https://www.argaam.com

UAE market: https://www.argaam.com/ae-ar

The same workflow must work on both regions (with region-aware paths/selectors where needed).

What the scraper must do for every ticker
1) Overview page (save + scrape tables)

Save the raw overview HTML

Scrape these tables from the overview page:

Market Data table

Financial Ratios table

Output tables in CSV (and/or JSON), clearly labeled per ticker + market

2) News (6 months)

From the overview page, click “More” under News and:

Collect all news articles going back at least 6 months

Save each full article (HTML) and store metadata (title, date, URL)

Avoid duplicates; handle pagination correctly

3) Earnings (3 years)

From the overview page, click “More” under Earnings and:

Collect all earnings-related articles going back 3 years

Save full article HTML + metadata per item

4) Disclosures (2 years)

From the overview page, click “More” under Disclosures and:

Collect all disclosure articles going back 2 years

Save full article HTML + metadata per item

5) Events (from Overview → Events table → More)

From the overview page, in the Events table click “More” and:

Scrape the events listing (and details if they open as separate pages)

Save HTML + structured output (CSV/JSON) with fields such as date, event type/title, and any linked detail URL

Handle pagination if applicable

Left-side tabs to scrape (after overview)
6) Profile tab (page + articles if any)

Save the profile page HTML

Scrape key tables/data on profile (at minimum full HTML reliably)

If the Profile tab contains articles/updates, scrape them as well:

Capture article listings + follow and save each full article page where applicable

Store metadata (title/date/url) + deduplicate

7) Chart tab → Calendar date range → Price history (10 years)

Go to the chart page

Use the calendar/date range UI:

Set From = 10 years back

Set To = today (default is fine)

Click Search

Wait for the long table to load

Scrape the resulting full price history table and save it as CSV

8) Financial Statements tab (Quarter + Annual + PDF downloads)

Page loads in Quarter view:

Save page HTML

Download the 5 English PDFs (buttons labeled EN)

Switch to Annual view:

Save page HTML

Download the 5 English PDFs again (EN)

Store PDFs in clearly labeled folders per ticker and period (quarter/annual)

9) Financial Result Chart tab

Sometimes opens in a new tab/window; handle that reliably

Save the page HTML (screenshots optional, but useful)

10) Investor Presentations tab (2 years, EN PDFs)

Collect and download all investor presentation PDFs from the last 2 years

English only where available (EN)

Save metadata (title/date/url) + downloaded PDFs

11) Corporate Actions tab → Dividend History + Capital Changes

Scrape the table for:

Dividend History (default selected tab)

Switch to Capital Changes tab and scrape that table too

Save both tables to CSV (and/or JSON) per ticker

Output & folder structure requirements

Output must be organized per ticker under a single project/universe folder.

For each ticker + market, I expect:

Raw HTML dumps (per section)

CSV exports (tables / listings / price history / corporate actions / events)

Downloaded PDFs in labeled folders

A meta.json (or similar manifest) recording:

run timestamp

market/region (KSA or UAE)

login status

counts of items scraped/downloaded per section

errors/warnings with debug screenshots/HTML on failure

Login/session requirements

Use a persistent Playwright profile so login persists between runs.

First run may require manual login + 2FA (acceptable).

Script must detect logged-in vs logged-out state reliably before scraping.

Do not hardcode credentials in code. Use environment variables and/or persistent session storage.

Tech requirements

Python 3.x

Playwright (sync or async is fine)

BeautifulSoup is fine for parsing HTML

Must run on a Linux server/VM environment

Strong preference for:

robust selectors (CSS/XPath) and fallbacks

retry logic + timeouts

pagination correctness

deduplication

clear logs

What I will provide

My current script (draft)

Ticker list + market mapping (KSA/UAE) and initial URLs/paths

Sample outputs and run logs

If needed, I can provide:

saved HTML files

screenshots of each target section

a single “golden ticker” test case for validation

Deliverables (definition of done)

Working scraper that runs end-to-end for one ticker in KSA and one ticker in UAE.

Correct outputs for each required section above (including Events and Profile articles if present).

Stability improvements:

handles popups/login overlays

handles new tabs/windows

handles pagination and date filters reliably

Documentation:

how to run

required environment variables

troubleshooting notes

How to apply (please answer these)

Share 2–3 examples of Playwright scrapers you built (login + pagination + downloads + date pickers).

Explain your approach for:

persistent login sessions + 2FA

date-range UI interaction for 10-year tables

reliable PDF downloads

deduplication strategy across multiple sections

Confirm you can support both KSA (argaam.com) and UAE (argaam.com/ae-ar) versions.

Propose a milestone plan (I prefer milestone-based delivery).

Preferred milestone structure

Login/session + overview tables + discovery (KSA + UAE)

News/Earnings/Disclosures + Events (pagination + full capture)

Profile tab + profile articles capture

Chart 10y price history extraction

Financial statements + PDFs, investor presentations, result chart

Corporate actions + final QA on multiple tickers + documentation

If you need any specific artifacts (HTML dumps, screenshots, or a single ticker test case) to speed up implementation, tell me exactly what you need and I will provide it immediately.

Absolute Requirements

If you do NOT have verifiable, production experience in ALL of the following, please DO NOT APPLY:

Building a real-time streaming voice pipeline (Twilio → ASR → LLM → TTS → caller).
Configuring a Twilio SIP trunk to an AWS External Load Balancer.
Building multiple microservices, dockerizing them, and deploying to ECS.
Producing and consuming AWS SNS/SQS events in a distributed system.
Implementing barge-in and real-time interruption handling.
Integrating LiveKit for realtime call answering agents

If you cannot prove (code samples, repos, references) that you’ve done this before, please do not apply.

We are building a production-grade Voice AI system with five independently deployable services (Inbound Agent, Outbound Agent, Listener, Classifier, Messaging Service). We need a senior, verifiably experienced engineer who has built real-time streaming voice pipelines, dockerized them, and deployed them into AWS ECS in an event-driven architecture.

• Tell me about the specific project where you first built a real-time voice or telephony-based AI system. If you use AI to answer, you will be disqualified
• Walk me through a project where you integrated Twilio SIP trunking into a backend service or microservice architecture. If you use AI to answer, you will be disqualified
• Describe a project where you built multiple microservices that communicated through a message queue like SQS/SNS. If you use AI to answer, you will be disqualified
• Tell me about a project where you worked with real-time audio streaming — LiveKit - If you use AI to answer, you will be disqualified
• Can you describe a project where you containerized multiple backend services and deployed them to AWS ECS or Kubernetes? If you use AI to answer, you will be disqualified

RFP: Web-Based Image Overlay & Alignment Tool (QGIS-Powered)
Overview
I am seeking a developer or small team to build a turn-key web application (SaaS) that performs a very specific, repeatable workflow for aligning raster images (PNG orthos) onto a reference PDF floor plan, using a simple, guided, step-by-step interface.
This tool is intended for non-GIS users and must not expose the full QGIS interface or complexity, even though it should use QGIS libraries/engine on the backend for reliability and accuracy.
The target audience includes users of NavVis VLX who frequently need to overlay orthographic imagery onto floor plans in order to validate whether a job site has been completely scanned or has “gaps” in scan coverage – prior to demobilizing from the site.
________________________________________
Core User Workflow (Must Be Followed Exactly)
Step 1 — Upload
• User uploads:
o One reference PDF floor plan (single page or page-selectable)
o One or more PNG overlay images
• The system converts the PDF page into a raster image for processing.
• All processing is done in pixel space (no real-world coordinates or GIS CRS required).
Step 2 — Align Images (Repeat per Image)
For each PNG overlay image, the user performs:
1. The interface displays split screen:
o PDF on one side
o PNG on the other side
2. User is prompted to:
o Click Point 1 on the PNG
o Click the same Point 1 on the PDF
o Click Point 2 on the PNG
o Click the same Point 2 on the PDF
3. The system applies a 2-point Helmert (similarity) transformation:
o Translation
o Rotation
o Uniform scale
4. The aligned image is immediately previewed on top of the PDF.
5. User can adjust opacity and either:
o Accept the alignment, or
o Re-pick points and try again
Step 3 — Optional Per-Image Mask / Clip
• For each aligned image, user may optionally:
o Draw a polygon, rectangle, or freehand shape
o This shape is used to clip ONLY that overlay image
• The mask is not applied to the final PDF, only to that specific overlay.
Step 4 — Export
• Final output:
o A PDF with all accepted overlays composited on the floor plan (PNG, JPG, TIFF are acceptable export formats)
• Optional secondary outputs (nice-to-have):
o Individual masked overlay images (PNG/TIFF)
o High-resolution raster export
________________________________________
Technical Requirements (What the Developer Needs to Know)
Backend
• Must use QGIS libraries / engine (headless):
o PyQGIS or qgis_process
o QGIS used strictly for:
 Raster warping
 Helmert transformation
 Raster clipping
 Raster compositing
• Processing is pixel-based only
o No CRS
o No georeferencing
o No map projections
Frontend
• Browser-based UI (no desktop installs)
• Purpose-built, guided workflow:
o Step-by-step prompts
o No layer trees, GIS tools, or advanced settings
• Must support:
o Point picking on PDF and image
o Polygon drawing for masks
o Live preview with opacity control
Architecture
• Web API (REST)
• Background job processing (image operations may be large)
• File storage for uploads and outputs
• Should be deployable via containers (Docker preferred)
________________________________________
Non-Goals (Explicitly Out of Scope)
• Full GIS editing
• Coordinate systems
• Georeferencing to real-world locations
• Manual layer management
• QGIS Desktop plugins
• User-facing access to QGIS UI
________________________________________
Success Criteria (How “Done” Is Defined)
• A non-technical user can:
o Upload a PDF
o Align multiple images using only 2 clicks per image
o Optionally mask each image
o Export a clean overlay PDF
• No GIS knowledge required
• No local software installs
• Consistent, repeatable results
________________________________________
Licensing & Distribution
• The application will be:
o Hosted as a SaaS
o Shared with a broader professional community
• Developer should be aware of QGIS GPL licensing implications for a hosted service.
________________________________________
What to Include in Your Proposal
Please include:
1. Relevant experience (GIS, QGIS, raster/image processing, web apps)
2. Recommended tech stack
3. High-level system architecture
4. Estimated timeline & cost range
5. Examples of similar tools (if available)
________________________________________
Optional Enhancements (Not Required, But Valuable)
• Ability to add a 3rd control point (advanced mode)
• Alignment quality feedback
• Undo/redo for point selection
• Auto-delete files after a retention period
________________________________________
Summary (Plain English)
I want a simple web tool that lets users line up images on a floor plan by clicking two matching points, optionally trimming each image, and exporting a clean PDF — powered by QGIS behind the scenes, without exposing any GIS complexity.

1. I have code that downloads daily and minute data from IB its in file.
I have also included sample parqet files downloaded. If you already have better code for download data that you want/can use, it is up to you. If it makes it easy for you.

2. I will have all s and p 500 stocks in daily_tickers.txt. So your code should be able to run simple stock selecting strategy like some technical indicators, momentum/mean rev that will short list few stocks to buy or sell next intraday. This is just to reduce load on minute data strategy. So we can only concentrate on few stocks..

3. We should have intraday strategy that will get trading signals using similar momentum/mean rev and we trade. Number of stocks we buy/sell can be fixed number of shares or dollar ammount.

4. We should be able run backtest and live trade both strategies

We are seeking a developer who has an understanding of base44 and lovable to tackle challenges. We have created a couple projects using base44 and lovable. Your primary responsibility will be to move the product to production while addressing some existing bugs. Someone who is detail-oriented and can ensure a smooth transition and functionality of the product. We would love to hear from you!

• Any questions? We would like to ensure no misunderstanding
• Describe your recent experience with similar projects
• Include a link to your GitHub profile and/or website
• What frameworks have you worked with?

1) anti-robot captcha
There is already ready-made software , its using MITM, to detect links. It works, but sites with anti-robot captcha are not avaliable because its detecting my pc as not human, so captcha cant be solved at all even by me by clicking on it with mouse(it detects pc like a bot), thats not possible to open any service where there is a captcha because its impossible to pass it.
Also without script working captcha is not even showing me on needed websites.

2. Issue with library .NET (screenshot 1)
issue with .NET while openning application.

Maybe rewritting of the script will be needed on other language(python/c++).

I am seeking a skilled Python developer having experience with PYQT6, ONNXRUNTIME and pyinstaller to troubleshoot and resolve a DLL load failure issue while importing 'onnxruntime_pybind11_state'. I have already built an application that can be deployed but the library rembg is not properly imported (see error above). It prevents me to successfully remove the background of the images that I load.

I am looking for a technology professional who is interested in teaching. Suitable for someone with industry experience and holds relevant experience in developing AI apps using python. Can explain NLP, CV, DL concepts with practical examples using python.

Following book can be used:
https://www.amazon.com.au/Learn-AI-Python-techniques-Scikit-Learn-ebook/dp/B09JP6QKRJ

Again the intent here is not to go deep in theory, rather cover the key concepts quickly with practical use cases/examples to get the intuition.

Budget: $12-$15 per hour
Duration: approx. 20 hours (2 hours per chapter from the above book, 1 hr theory + 1 hr practical) to be covered within 2 weeks

We are seeking a developer to create a Product Stock Tracking System utilizing Binary Search Trees (BST), preferably using Java. This is a homework project, so it is essential that all work is original and that AI tools are not used. The system should efficiently manage and track product stock using BST principles. Candidates must have a solid understanding of data structures, particularly Binary Search Trees, and be capable of implementing them in Java. Please provide examples of previous work related to data structures.

**Description**

We are seeking a highly skilled AI Developer to fine-tune or optimize a legal reasoning/chatbot tool using a structured legal dataset.

Your goal is to meet or exceed a predefined accuracy threshold on legal question-answering benchmarks, including ruling prediction, statute interpretation, and citation correctness. This is a results-based contract with a hard deadline.

You will be provided:

* A cleaned, structured legal dataset (JSONL format)
* Evaluation scripts and scoring logic
* Infrastructure ready to support fine-tuning or RAG (no setup needed)
* Support developer for preprocessing and technical integration

---

**Responsibilities**

* Fine-tune an open-source LLM (e.g., Mistral, LLaMA, Yi, DeepSeek) or implement a RAG pipeline
* Optimize for accuracy, factual consistency, and reduced hallucination
* Evaluate results using provided metrics and benchmarks
* Deliver output that meets or exceeds the defined accuracy goal

---

**Requirements**

* Experience fine-tuning LLMs or building high-accuracy RAG systems
* Familiarity with Hugging Face Transformers, FAISS, PyTorch, and quantization
* Ability to deliver measurable improvements and meet strict deadlines
* Strong results-oriented mindset
* Ph.D. or Master’s degree REQUIRED in Artificial Intelligence, Machine Learning, Computer Science, or a closely related field

---

**Budget**

* Fixed: $1,000 base
* Plus up to $1,000 performance bonus

Bonus is awarded if accuracy exceeds the defined threshold by +5% or more.

Payments shall be issued in two installments:
(i) Initial Payment: $500 released upon delivery of a working fine-tuned or RAG-based model and functioning inference script
(ii) Final Payment: $500 released only after verification of all deliverables, accuracy benchmarks, documentation, and completion of the offboarding process

A performance bonus of up to $1,000 may be awarded if the Contractor exceeds the accuracy threshold (90%) by five percent (5%) or more, as verified by HQ.

---

**Timeline**

21-day deadline, start to finish — no delays. This project is critical and time-bound.

---

**To Apply**

Please include:

* Links to previous LLM fine-tuning, QA chatbot, or RAG-related projects
* Confirmation that you can complete the project within 21 days
* Any questions you have about the data or evaluation process

---

**Tags (Skills)**

LLM, Fine-tuning, PyTorch, Hugging Face, FAISS, RAG, GPT-style models, AI Model Optimization, Machine Learning, Natural Language Processing (NLP), LegalTech

---

**Deliverables**

1. Fine-tuned or Optimized Legal Chatbot Model

* Either via direct fine-tuning of an open-source LLM or via a RAG pipeline
* Model must handle statutes, rulings, citations, and general legal QA

2. Model Integration Script(s)

* Scripts to load and run the model on our infrastructure (Dockerized or Python-based)
* Must include an inference entry point such as predict(question)

3. Evaluation Report

* Accuracy performance against benchmark set (provided)
* Before/After metrics with exact percentage changes
* Explanation of how accuracy was achieved

4. Bonus Threshold Summary (only if eligible)

* Report showing threshold exceeded by +5%
* Justification for performance bonus eligibility

5. All Code Files

* Training scripts, configuration files, model files, and integration code
* Clear structure and documentation for internal use and maintenance

6. Deployment Instructions

* Step-by-step instructions for running inference on our server
* GPU usage notes and environment file

7. One Live Demo (optional but preferred)

* Short video or screenshare showing the model handling test prompts

• Describe your recent experience with similar projects

1. Coupon generation & storage
1. Landing page / form:
o Create a simple Google Form or Typeform linked to a QR code. Required fields: Full name, phone number, email, event name (e.g. “XYZ Event”). Add a privacy consent checkbox.
o Separately, prepare a Google Sheet “Event Coupons” with columns: Full Name, Phone, Email, Event, Coupon Code, Valid Until, Used?, Used By, Hotel Name.
2. n8n workflow (triggered on form submission):
o Trigger: Use n8n’s Google Forms or webhook node to capture each submission.
o Generate coupon: Create a unique code (e.g. BT-random5). Set Valid Until to 31 December 2026 (or two months after issue if you prefer).
o Record to Sheets: Append a new row to the “Event Coupons” sheet with the attendee’s details, the code, the expiry date, and Used? = NO.
o Create/update contact in Pipedrive: Use the Pipedrive API node to search by email/phone; if no contact exists, create a new Person with the attendee’s info and attach the coupon code as a custom field or note.
o Send coupon to attendee: Use n8n’s email node to send an email (and/or WhatsApp via your Business API) thanking them for attending and providing the 10 % discount code, along with redemption instructions and the expiry date.
3. Bulk attendee list from organiser: If the organiser provides a list of attendees after the event, put it into the same Google Sheet and run the same workflow in bulk (loop through each row → generate code → update sheet and Pipedrive → send email).
________________________________________




2. Coupon redemption via WhatsApp
1. WhatsApp incoming message handling:
o Set up n8n to receive WhatsApp messages via your WhatsApp Business API webhook.
o When a message arrives, respond with a menu (ChatGPT can help craft this):
1. Chat to a human
2. Redeem discount code
3. Plan my trip
2. Redeem code flow:
o If the user chooses option 2, ask which event they attended (list out “AUB Nurses Event” and any others).
o Once the event is selected, ask for either their coupon code or their name/email/phone to look it up.
o Search the “Event Coupons” sheet for a matching record where:
 Event matches,
 Used? = NO,
 Coupon code or contact details match the user’s input.
o If a match is found:
 Confirm validity: “Your 10 % discount is valid! I’ll transfer you to a human agent to assist with booking.”
 Create a Pipedrive activity for an agent to follow up.
o If no match is found or the code is already marked Used, send a polite message explaining that the code is invalid or has already been used, and offer to connect them to a human for help.
3. Mark coupon as used:
o When the booking is completed, have the agent or an automated process update the row in the sheet:
 Used? = YES,
 Used By = customer name,
 Hotel Name, Price Online, Price After Discount.
o Optionally, update the corresponding Person in Pipedrive with a note indicating the coupon was redeemed.
________________________________________
Key integration points
• Pipedrive integration: Every new coupon signup triggers a person search in Pipedrive; create or update contacts accordingly so you have a complete attendee database.
• Validation logic in n8n: The WhatsApp redemption flow only checks codes against the sheet; if valid and unused, it proceeds, otherwise it blocks redemption.
• Security: Use strong random coupon codes and restrict each code to one use by marking it “Used” after redemption. Store coupon expiry dates and enforce them when validating.
• Expiry: For the AUB nurses event, set all generated codes to expire on 31 December 2026. Adjust the “Valid Until” field as needed for other events.

Looking for rfid specialist who has experience
Zebra, Chainway etc
Wms is built laravel vue

**Job Description:**

Seeking a skilled developer to patch my Home Depot web scraper. Recent anti scraping protections have prevented my script from functioning. Candidates should have knowledge of how to get around these measures.

I will provide My current script, with associated files.

Project Overview

I am looking for an experienced developer to build a robust, rule-based email parsing utility that extracts structured data from two Outlook .pst files and outputs normalized relational tables (CSV).

This is a local PST parsing and ETL task, not web scraping and not UI work.

The PST files contain two categories of emails, stored in separate folders:

Approval emails

Request-for-approval emails

Each category has multiple known formatting variants (same business meaning, different layouts/wording).

Project Structure & Milestones
Phase 1 (Initial Milestone — $1,200 fixed price)

Parse PST files and extract data from:

5 approval email variants

5 request-for-approval email variants

Produce validated CSV outputs for all tables below

Implement required-field validation and error reporting

Phase 2+ (Incremental Variants)

Add support for additional email variants at:

$75 per variant

Variants will be added in small batches, using sample emails I provide

Data Model (Normalized)
Approvals

ApprovalsHeader

ApprovalID (PK)

ApprovalDate

Company

NumberOfFilesReviewed

NumberOfTransfersRequested

(optional: SourceMessageID)

ApprovalsFiles

ApprovalFileID (PK)

ApprovalID (FK)

FileReviewed

ApprovalsTransfers

ApprovalTransferID (PK)

ApprovalID (FK)

LoanX_YN

DebtEquity

LoanEquityType

Price

GrossAmount

NetAmount

GCtoEntity

GCfromEntity

Requests

RequestsHeader

RequestID (PK)

RequestDate

ApproveByDate

Company

NumberOfTransfersRequested

(optional: SourceMessageID)

RequestsTransfers

RequestTransferID (PK)

RequestID (FK)

LoanX_YN

DebtEquity

LoanEquityType

Price

GrossAmount

NetAmount

GCtoEntity

GCfromEntity

Validation & Error Handling (Required)

Required fields must be validated per email type

If required data is missing:

The email must be logged to an Errors/Rejects output

Include message identifier (date, subject, sender, internal ID)

Include which required fields were missing

Script must continue processing remaining emails

Inputs Provided

I will provide:

Sample PST file for approvals

Sample PST file for requests

Example emails for each supported variant

Field definitions and clarification as needed

Deliverables

Script (Python preferred; C#/.NET acceptable)

CSV files for each output table

Error/Rejects CSV

Brief documentation describing:

How variants are handled

How IDs are generated

How to rerun the script on new PSTs

Required Experience

Prior work parsing Outlook PST files (libpff / pypff / Outlook Interop / MAPI)

Strong experience with email ETL / text parsing

Regex and rule-based extraction from unstructured content

Designing normalized data outputs

Comfortable handling real-world “messy” emails

Please do not apply if you have not worked on PST or email-parsing projects before.

How to Apply

Please include:

Brief description of similar email or PST parsing projects you’ve completed

Language/tools you would use and why

Confirmation you are comfortable with incremental variant-based pricing

Additional Notes

No UI required

No Outlook automation required

No cloud deployment

This is a deterministic parsing task, not NLP

SUMMARY
Build a secure, production-ready web app + browser extension that helps authors manage journal manuscript submissions across fragmented editorial platforms (ScholarOne, Editorial Manager, Taylor & Francis, etc.) without any journal/publisher integration. Core value: one dashboard + one-click deep links + correct per-journal credential autofill, plus reminders and optional email-based updates.

This is an end-to-end build: architecture, implementation, security, testing, deployment, and documentation. UI/UX should be clean and upgradeable (room for improvement and iteration), but the focus is functional reliability and trust.

Suggested Tech Stack ( may propose alternatives)
• Frontend: React / Next.js
• Backend: Node/NestJS or Python/FastAPI
• DB: Postgres (Supabase acceptable)
• Auth: Supabase Auth / Auth0 / custom
• Hosting: Vercel/Fly/Render + managed Postgres
• Email: SendGrid/Mailgun
• Payments: Stripe
• Extension: MV3 + TypeScript
Security expectations:
• HTTPS everywhere
• Encrypted secrets
• Password hashing (Argon2/bcrypt)
• RBAC and tenant isolation
• OWASP baseline hardening
• Backups + disaster recovery plan

Deliverables (partial list)
1. Production web app (responsive)
2. Chrome extension published (or ready-to-publish package)
3. Backend/API with documentation
4. Database schema + migrations
5. Stripe billing live
6. Security documentation:
- Threat model
- credential storage design
- data retention policy
7. QA:
o test coverage for critical paths
o manual test checklist
8. Deployment:
o CI/CD pipeline
o monitoring/logging
9. Handoff:
o runbook for operations
o onboarding docs
o “how to add support for new platform patterns”

Will provide full Job Description upon request to better assist with providing a quote and timeline.

We are seeking a highly skilled software engineer with a strong background in AI and Google DeepMind technologies. The ideal candidate will possess exceptional creativity and the ability to design cutting-edge AI models that push the boundaries of innovation. If you have a passion for pioneering advancements in artificial intelligence and a proven track record in developing frontier models, we would love to hear from you. Join us in shaping the future of technology!

• Describe your recent experience with similar projects
• Please list any certifications related to this project
• What tools do you use for data mining and visualization?
• How do you deal with outliers or missing values in a dataset?
• What techniques would you use to clean a data set?

We are seeking an experienced backend engineer to help design and build a production-grade, AI-enabled document intelligence platform. This system will ingest large volumes of heterogeneous documents (PDFs, images, structured and unstructured files), process them through multiple analysis stages, and expose search, retrieval, and analytical capabilities via APIs.

This is not a prototype or academic exercise. The work is focused on building scalable, reliable, and maintainable systems suitable for regulated or enterprise environments.

You will collaborate closely with a technical lead who provides architecture direction, requirements, and review. The expectation is disciplined engineering execution, not exploratory tinkering.

Responsibilities

Design and implement backend services for document ingestion, processing, and analysis

Build API-driven services (REST/JSON) with clear contracts and strong validation

Implement pipelines for OCR, document parsing, metadata extraction, and AI-based enrichment

Integrate vector search and/or semantic retrieval capabilities

Work with structured and semi-structured data models (JSON, relational schemas, embeddings)

Optimize for performance, throughput, and operational reliability

Write clean, testable, production-ready code with appropriate logging and error handling

Collaborate on deployment and runtime considerations (containers, service configuration)

Required Skills & Experience

Strong Python backend development experience (FastAPI or similar frameworks)

Experience building API-first systems and microservices

Hands-on experience with document processing (PDFs, OCR, image pipelines)

Familiarity with AI/ML integration (LLMs, embeddings, classification or extraction tasks)

Experience with search or retrieval systems (e.g., Elasticsearch, OpenSearch, vector databases)

Solid understanding of data modeling and schema evolution

Comfortable working with Docker and containerized services

Ability to work independently with clear communication and accountability

Nice to Have

Experience with cloud-native architectures (Azure, AWS, or GCP)

Familiarity with event-driven or workflow-based systems

Knowledge of security and access control in enterprise applications

Prior work in legal, compliance, investigative, or regulated data domains

Engagement Details

Contract role with potential for long-term engagement

Flexible hours, but availability for regular technical check-ins is required

Immediate start preferred

Competitive hourly or milestone-based compensation, commensurate with experience

How to Apply

Please include:

A brief summary of relevant experience building backend or AI-enabled systems

Examples of similar platforms or pipelines you’ve worked on (no confidential details)

Your preferred engagement model (hourly vs. milestone) and availability

Odoo Delivery Note : customize to include customer balance, drivers name, Vehicle, customer and driver signature. Delivery note should also be printable only once
Dashboard with many datapoints and KPIs
Create a solid backup plan for the system
this should take a few hours.

Hello,
I’m looking for an experienced developer to build a semi-automated Etsy listing system that is fully compliant with Etsy policies.

The goal is to speed up listing creation, not to violate Etsy rules or bypass safeguards.

✅ CORE REQUIREMENTS

1. AI Image Generation Pipeline

Generate product images using AI (Midjourney / Stable Diffusion / DALL·E or similar)

Support prompt variation to avoid repetitive visuals

Images must look natural, realistic, and suitable for Etsy

No copyrighted characters or restricted content

2. AI Listing Content Generator

Automatically generate:

SEO-optimized title (Etsy best practices)

13 tags

Category & attributes

Product description:

SEO intro

Bullet points

Care / delivery info

Content must be unique per listing

No keyword stuffing or spammy behavior

3. Etsy Upload (SAFE & COMPLIANT)

Use Etsy official API only

Respect rate limits

No scraping, no browser automation (Selenium/Playwright)

Prefer manual approval step before publishing

4. Human-Like Workflow

Designed for:

1–2 Etsy shops

Low daily volume (1–5 listings/day)

NOT mass posting

NOT multi-account abuse

NOT automation meant to bypass Etsy systems

⚠️ IMPORTANT COMPLIANCE RULES

This project must strictly follow:

Etsy Seller Policy

Etsy API Terms of Use

No account creation automation

No IP rotation, fingerprint masking, or evasion logic

No fake shipping or misleading listings

If any feature risks account suspension, it should not be implemented.

🛠️ TECH PREFERENCES

Backend: Python or Node.js

Optional dashboard (simple web UI)

Clean, documented code

Modular structure (image → content → upload)

📦 DELIVERABLES

Working MVP

Setup instructions

Code documentation

Basic error handling & logs

⏱️ ESTIMATED SCOPE

MVP: 20–40 hours

Future improvements possible after testing

Please include in your proposal:

Similar automation or marketplace experience

Tech stack suggestion

Estimated timeline

Any compliance risks you foresee

-Umut