Discover Computer Vision Projects Efficiently

We are seeking a skilled developer or engineer to design and implement a vision-based system for counting tablets and capsules, replacing our existing IR system. The ideal candidate will have experience in computer vision, image processing, and automation solutions. Your role will involve developing software that integrates with our production line, ensuring accuracy and efficiency in counting. If you have a strong background in this field and can deliver innovative solutions, we would love to hear from you.

• Describe your recent experience with similar projects

Need a computer vision developer (Python, OpenCV, MediaPipe, or similar)
• Build a lightweight real-time vision pipeline
• Detect faces from a webcam or phone camera
• Detect a visible marker (printed symbol or pattern) in the scene
• Apply conditional image filtering (blur/mask) when marker is present
• Focus is on proof-of-feasibility, not a commercial product
• No facial recognition, identity tracking, or databases involved
• Concept is patent pending, full details shared after initial discussion

About the Project
We're looking for an experienced Android developer to help build AI-powered mobile applications. The work involves integrating machine learning models for real-time detection and analysis, both running locally on-device (offline) and via cloud APIs.
Note: Your task will be integration of AI models in app architecture but its nice to have knowledge about AI models.

What You'll Be Working On
Integrating custom detection models (YOLO, TensorFlow Lite, ONNX) into Android apps
Implementing on-device inference for offline functionality
Connecting apps to AI APIs for cloud-based processing
Optimizing model performance for android mobils(Snapdragon 6 cpus etc.)
Building clean UI to display detection results in real-time
Required Skills
Strong Android development experience (Kotlin/Java/Flutter)
Hands-on experience with ML model integration (TFLite, ONNX Runtime, MediaPipe, or similar)
Experience with REST APIs and handling async operations
Understanding of camera integration and real-time video processing
Nice to Have
Experience with sports or video analysis apps
Background in computer vision
Published apps with AI/ML features
Test Task ($50 Paid)
Before starting the main project, there's a short paid test task: integrate a provided detection model into a basic Android app with real-time camera preview. This helps us assess your skills and working style.
To Apply
Share relevant projects or apps where you've integrated AI/ML
Briefly describe your experience with on-device model deployment
Mention your availability and hourly rate
Looking forward to working with a skilled developer who's excited about AI on mobile!

In your reply add the following:
1. Your Github
2.Past projects
3.Why are you a good fit for this project.

I’m a kitchen designer with 20+ years of experience and 1,000+ completed kitchen designs. I want to build an AI “junior designer” workflow.

Goal: I upload a floor plan DXF (and optionally a few reference images) + a short brief (colors, textures, appliances, preferred heights, island yes/no, style), and the system generates photoreal, high-end kitchen render views for that specific space.

I prefer a freelancer who can propose and implement a solution using existing, proven AI models/workflows (e.g., Stable Diffusion/SDXL + ControlNet + LoRA/DreamBooth or equivalent), rather than inventing a model from scratch.

⸻

What the Tool Must Do (Core Requirements)

Inputs
• DXF floor plan (I can prepare/export this cleanly)
• Project requirements such as:
• cabinet colors + textures/material style
• countertop type
• flooring/wall finish direction
• appliance selection (brand/style: e.g., built-in fridge, oven tower, cooktop type)
• preferred kitchen height standards (base/wall/tall logic)
• island yes/no, approximate size/shape preference
• overall style: modern/luxury/minimal, etc.

Outputs
• 3–6 photoreal render views (day + warm interior options preferred)
• High resolution (target 4K, with upscale option)
• Consistent style (high-end interior photography look)

Accuracy constraint (very important)

The system must preserve the room geometry and layout from the plan:
• straight lines stay straight (no warped cabinets/walls)
• doors/windows remain in correct locations
• the layout should respect space constraints

(You can use structure guidance methods like ControlNet edge/depth/lineart or equivalent to lock geometry.)

⸻

Preferred Approach (Not Starting From Scratch)

I want you to:
1. Propose an existing AI stack that already works for image generation
2. Add the ability to learn/adapt to my style using my dataset (fine-tuning like LoRA, style adapters, retrieval-based prompting, etc.)
3. Build a practical tool I can run (local PC with GPU or cloud)

I’m open to:
• Local pipeline (preferred if stable)
• Cloud pipeline (if cost is predictable and easy)

⸻

Deliverables
1. Proof of concept on 1–2 sample plans (show before/after and explain approach)
2. Working MVP tool (choose one):
• Desktop app / packaged script (Windows-friendly), OR
• Simple web UI (upload DXF → choose options → generate)
3. Style adaptation method:
• training/fine-tuning workflow
• ability to improve results over time by adding new examples
4. Documentation:
• install/run instructions
• hardware requirements
• per-image runtime estimate
• cost estimate if using cloud GPUs
5. Project files + source code + clear licensing guidance (commercial use)

⸻

Data I Will Provide
• Sample DXF floor plans (clean exports)
• Example “good” kitchen designs/renders (my style)
• A small starter dataset (e.g., 50–150 projects) for the first iteration
• Reference images for desired realism level (high-end interior photography)

Confidentiality: Client data must remain private. NDA is required.

⸻

Milestones (Suggested)

Milestone 1 — Discovery + POC (1–2 weeks)
• Confirm DXF ingestion method and best AI pipeline
• Generate 4–8 variants on 1 sample plan
• Show how structure is preserved (no warped geometry)

Milestone 2 — MVP Build (2–6 weeks)
• Parameter controls (materials/colors/appliances/island preference)
• Batch generation of multiple camera angles
• Save presets (styles)

Milestone 3 — Style Adaptation (2–6 weeks)
• Fine-tune/adapt to my dataset
• Compare “before vs after”
• Add feedback loop (“this result is good/bad” and improve)

Milestone 4 — Packaging + Handover
• Installer / packaged workflow
• Documentation
• Final testing on 10–20 sample plans

(You can propose a different plan if you have a better workflow.)

⸻

Skills Required
• Stable Diffusion / SDXL (or equivalent)
• ControlNet / structure guidance (edge/lineart/depth)
• LoRA / DreamBooth / style adaptation
• Python + building repeatable pipelines (ComfyUI or custom)
• Image upscaling / post-processing workflows
• Experience with architectural/interior image generation is a big plus
• DXF/CAD parsing experience (bonus)

⸻

Screening Questions (Answer in Your Proposal)
1. Share 2–3 examples of similar work (interiors/archviz, ControlNet, style fine-tuning, image-to-image pipelines).
2. What stack do you recommend and why (local vs cloud, SDXL vs other)?
3. How will you keep the plan geometry accurate (no warping)?
4. How will you handle appliance selection (consistency + realism)?
5. What is your estimated:
• time to deliver a working POC?
• time to deliver MVP?
• budget range (POC / MVP / full)?
6. Any licensing constraints for commercial use of the models/workflows you propose?

⸻

Budget / Contract Type

I’m open to:
• Fixed-price with milestones (preferred), OR
• Hourly for discovery + fixed for milestones

Please propose 2–3 options:
• POC only
• MVP
• Full version (style-adaptive + workflow polished)

⸻

Very important:

If you have an “already working” interior AI workflow (existing model + proven pipeline), please describe it—my preference is to build on something functional rather than reinvent everything.

Obfuscate and encrypt a c++ computer vision program whilst maintaining baseline performance. Project is not completely finished at the moment so consultation and input would also be appreciated.

• Describe your recent experience with similar projects

I need a small-footprint Python program that watches a live webcam feed on my desktop, recognises visible flames in real time with OpenCV, and immediately triggers two responses:

• a pop-up or full-screen visual warning on the PC;
• an automatic email alert sent to a configurable address.

Accuracy and low latency are more important to me than fancy UI work. The core of the job is a reliable flame-detection model, preferably using colour segmentation plus motion analysis or a lightweight CNN if that proves markedly better. All code must run locally under Windows 10 (Python 3.10, OpenCV 4.x).

Deliverables
1. Well-commented Python script(s) with clear setup instructions.
2. A requirements.txt for any external libraries.
3. Simple configuration file or section for camera index, SMTP credentials, recipient email, and alert text.
4. Short README that explains how to launch, test, and fine-tune sensitivity thresholds.

Acceptance will be based on a demo video or live screen-share showing:
• detection of an open flame within two seconds;
• visual warning appearing on screen;
• confirmation email received;
• false positives kept to a minimum when ordinary ambient light or people move through the frame.

If you have earlier work with OpenCV-based detection, especially fire or smoke, please mention it so I can gauge fit quickly.

Skills: Python, Windows Desktop, Linux, Software Architecture, Arduino, Coding, OpenCV, Computer Vision

We are seeking a skilled developer to create a personalized AI tutor that can adapt to individual learning needs and preferences. The ideal candidate will have experience in natural language processing, machine learning, and user interface design. Your role will involve designing the AI's learning algorithms, creating engaging content, and ensuring a smooth user experience. If you are passionate about education technology and have a proven track record in AI development, we would love to hear from you.

I need a simulation to test dual-arm robotic coordination using ROS2, Gazebo, Rviz, and MoveIt.

Key functional areas to cover:
- Object manipulation
- Path planning
- Collision avoidance

Ideal skills and experience:
- Proficiency in ROS2 and simulation tools
- Strong background in robotics, especially dual-arm systems
- Experience in developing and testing robotic algorithms in interbotix viperx300s

Skills: Wireless, Engineering, Matlab and Mathematica, Electrical Engineering, Robotics, Artificial Intelligence, Computer Vision, Simulation, Robot Operating System (ROS), Gazebo

I need a video tracking system to analyze the behavior of rodents for pharmacological research. The system should focus on tracking movement patterns.

Key Requirements:
- High accuracy in tracking rodent movement
- Real-time data processing and analysis
- User-friendly interface for monitoring

Ideal Skills:
- Experience in video tracking technology
- Background in animal behavior research
- Proficiency in data analysis software

Please provide a portfolio with relevant projects.

Skills: Data Processing, Software Development, Statistical Analysis, Data Analysis, Computer Vision, Machine Learning Algorithms

AI Development / Computer Vision / Full Stack
We are seeking a high-level Full-Stack Engineer to develop a modular authentication prototype. The objective is to synchronize three distinct data processing streams into a unified verification output with minimal latency.

Technical Requirements:

Stream 1 (Spatial Data): Implement real-time 468-point facial landmark tracking and eye-tracking using Google MediaPipe.

Stream 2 (Signal Extraction): Implementation of an RGB-based Signal Extraction library (e.g., rPPG) to monitor dermal pixel variations from standard video feeds.

Stream 3 (Pattern Logic): Development of a detector for custom, stochastic 2D dot arrangements using OpenCV.

Integration: Orchestrate these streams with an external verification API and commit finalized hashes to a Layer-2 Ledger (Polygon) via Thirdweb SDK.

Key Deliverables:

A clean, modular codebase (Python or Node.js).

High-performance synchronization of CV modules.

Comprehensive documentation for internal hand-off.

I’m commissioning a complete set of drone models that fuse onboard AI with true swarm behaviour for defence missions. The deliverable spans the airframe design, the electronics layout, and the software stack that lets multiple units cooperate autonomously in perimeter patrol, threat detection, and rapid interception scenarios.

A solid proposal will walk me through:

• Your preferred toolchain (e.g., SolidWorks or Fusion 360 for CAD, ROS 2 + PX4/Gazebo or similar for simulation and control).
• How you’ll implement real-time computer vision, mesh networking, and distributed decision-making without relying on constant ground-station oversight.
• A milestone plan demonstrating concept design, virtual flight-test, and (if practical within the month) a lightweight working prototype.

I’ll supply my threat-profile documents and performance targets on award. Please send a detailed project proposal outlining your methodology, timeline, and any prior work that proves you can take an idea from paper to coordinated, AI-driven flight inside the four-week window.

Skills: CAD/CAM, Solidworks, Manufacturing Design, Product Design, 3D Modelling, Drone Photography, AI Development, AI Agents

I’m looking to bring on a PhD-level researcher for a research project focused on deep learning, computer vision, and explainable AI. The role involves applied research and technical problem-solving.

Senior AI/ML Engineer - Real-Time Sales Copilot System (Python, OpenAI, Vector Search)

THE CHALLENGE

We've built a working AI-powered sales copilot that listens to live phone calls and provides real-time response recommendations to operators. The system works. Now we need an expert to make it exceptional—faster, smarter, and more contextually aware.

This isn't a build-from-scratch project. You'll inherit a functional codebase with a FastAPI backend, multiple production UIs, semantic search infrastructure, and a dataset of 25,000+ transcribed sales calls with extracted Q&A patterns. Your job is to optimize the AI pipeline so responses are highly relevant and delivered in near real-time as conversations unfold.

WHAT EXISTS TODAY

The current system includes a FastAPI server with WebSocket support for real-time streaming, a semantic matching engine using OpenAI text-embedding-3-small with cosine similarity search, GPT-4o-mini integration for generating contextual responses in a trained voice/style, 12,000+ extracted delivery examples from top sales performers, conversation state tracking with stage detection and signal accumulation and technique tracking, multiple production UIs with real-time updates, and a modular STT adapter layer ready for Vonage, Deepgram, AssemblyAI, or OpenAI Realtime.

The infrastructure works. The bottleneck is making the AI recommendations more precise and the response latency tighter.

WHAT WE NEED

Primary focus areas include response relevance (the semantic search sometimes returns tangentially related examples and we need better retrieval, possibly hybrid search, re-ranking, or fine-tuned embeddings), latency optimization (the full pipeline from STT to embedding to vector search to LLM needs to hit sub-2-second response times consistently), context window management (better use of conversation history to avoid repetitive suggestions and surface stage-appropriate techniques), intent classification (current keyword plus embedding hybrid approach works but could be sharper), and Vonage integration (connect the STT pipeline to live Vonage calls with minimal latency overhead).

TECH STACK

You'll be working with Python 3.11+ including FastAPI, uvicorn, Pydantic, and asyncio. OpenAI API for embeddings and chat completions. Numpy for vector operations including cosine similarity and index management. WebSockets for real-time streaming. HTML/CSS/JavaScript frontends which are not your primary focus but you'll touch them. JSON and JSONL data formats for knowledge bases. Vonage Voice API for upcoming integration.

THE DATA

You'll have access to 25,000+ transcribed sales calls redacted for privacy, 12,400+ extracted delivery examples which are customer question to successful response pairs, curated fact cards organized by intent, an intent taxonomy with 50+ categories, and a product knowledge base with specs, pricing, and competitive positioning.

This is real production data, not synthetic. The patterns are proven—they come from calls that converted.

REQUIREMENTS

Must have production experience with OpenAI APIs for both embeddings and chat completions, deep understanding of vector similarity search (not just "I've used Pinecone"), strong async Python with FastAPI and asyncio patterns, experience optimizing ML inference latency, and ability to read and improve existing codebases rather than just greenfield projects.

Strong preference for experience with real-time audio and STT pipelines, familiarity with RAG architectures and retrieval optimization, background in conversational AI or sales and support tooling, and experience with Anthropic Claude API which we use in tooling.

Not required: frontend expertise (we handle UI), DevOps and deployment (we handle infrastructure), or training custom models from scratch (we're optimizing, not researching).

ENGAGEMENT STRUCTURE

This is an hourly, ongoing relationship. The copilot is one of several AI projects in our pipeline. We're looking for someone who can become our go-to AI engineer across multiple initiatives—not a one-and-done contractor.

You'll have full access to the codebase, documentation, and call data. We work async-friendly but expect regular communication and incremental progress.

SCREENING QUESTIONS

Please answer these in your proposal. Brief answers are fine—we're looking for signal, not essays.

1. We're using text-embedding-3-small for semantic search on conversational sales text. Retrieval relevance is inconsistent. What's your diagnosis framework, and what alternatives would you evaluate?

2. Our pipeline chains STT to embedding lookup to LLM completion. Total latency varies from 1.5s to 4s. Walk us through how you'd profile this and where you'd focus optimization efforts first.

3. Describe a project where you improved an existing AI/ML system's performance in terms of relevance, latency, or accuracy. What was the baseline, what did you change, and what was the measurable outcome?

TO APPLY

Skip the generic proposal. Answer the screening questions, link to relevant work such as GitHub, past projects, or a brief Loom walkthrough, and tell us your hourly rate. We'll respond to qualified candidates within 48 hours.

• We're using text-embedding-3-small for semantic search on conversational sales text. Retrieval relevance is inconsistent. What's your diagnosis framework, and what alternatives would you evaluate?
• Our pipeline chains STT to embedding lookup to LLM completion. Total latency varies from 1.5s to 4s. Walk us through how you'd profile this and where you'd focus optimization efforts first.
• Describe a project where you improved an existing AI/ML system's performance in terms of relevance, latency, or accuracy. What was the baseline, what did you change, and what was the measurable outcome?

Job Description

Looking for a developer to set up a private AI image generation studio for personal use.

Scope:

Train one SDXL LoRA using 25–30 photos

Deploy fast image generation via RunPod or Replicate

Create a simple private web page (password protected)

Add presets for lifestyle, UGC, and fashion shots

Auto start GPU on use and auto shut down when idle

Not needed:

No payments

No multi-user system

No mobile app

Deliverables:

Images generate in under 1 minute after initial model load

Face consistency across styles

System runs without weekly maintenance

Budget: $150 fixed
Timeline: 3–5 days

• Have you personally trained an SDXL LoRA and deployed it for private use (RunPod, Replicate, or similar)? Briefly describe.
• What GPU setup would you recommend to reliably generate images in under 1 minute and why?
• Will this system run fully under my accounts with no ongoing dependency on you?

We are seeking a skilled Machine Learning Engineer with a strong background in building and deploying image diffusion models. The ideal candidate will be responsible for both training and inference processes, ensuring high-quality output from the models. You should have a deep understanding of machine learning frameworks and a passion for innovative image processing techniques. If you have experience in this niche area and are excited about contributing to cutting-edge projects, we would love to hear from you.

• Describe your recent experience with similar projects

Project Overview

We completed Phase 1 technical validation.

This Phase 2 project is a small, focused proof-of-concept to build a depth-first volume estimation pipeline using iOS ARKit LiDAR / sceneDepth when available, with RGB as a fallback.

This is not a full product.
No UI polish. No marketplace. No scheduling. No payments.

The goal is to produce defensible volume bands with confidence, not perfect measurements.

⸻

Core Objective

Build a minimum pipeline that:
• Ingests a short iOS walkthrough (LiDAR when available)
• Estimates truck-fill bands:
• ¼, ½, ¾, Full
• Outputs:
• band estimate
• confidence score
• failure / low-confidence flags

We are building a truth engine, not a consumer app.

⸻

Scope (What You Will Build)

IN SCOPE
• iOS ARKit depth ingestion (LiDAR / sceneDepth)
• Basic multi-view aggregation
• Rough volume estimation (non-precise is OK)
• Confidence scoring logic
• Explicit failure detection (bad capture, bad geometry, etc.)
• Simple structured output (JSON or similar)

OUT OF SCOPE (Explicitly Not Needed)
• Polished UI / UX
• Marketplace features
• Scheduling or payments
• Customer accounts
• Full automation
• Training ML models at scale
• Android support (later phase)

⸻

Deliverables
• Working prototype (local or simple repo)
• Clear explanation of:
• assumptions
• failure modes
• confidence logic
• Example outputs from test captures
• Short README explaining how to run + limitations

⸻

Milestones & Payment Structure

Milestone 1 — Depth Capture & Grounding (Week 1)

Goal: Prove reliable depth ingestion and scene grounding
Deliverables:
• ARKit LiDAR capture working
• Ground plane alignment
• Basic scene scale sanity checks
• Sample depth visualization

Success = depth data is usable and stable
25% payment

⸻

Milestone 2 — Volume Estimation → Truck Bands (Week 2)

Goal: Turn depth into rough volume bands
Deliverables:
• Multi-view aggregation
• Rough volume computation
• Mapping to truck-fill bands (¼, ½, ¾, full)
• Sample test outputs

Success = bands are reasonable on good captures
35% payment

⸻

Milestone 3 — Confidence & Failure Logic (Week 3)

Goal: Make the system honest
Deliverables:
• Confidence scoring logic
• Explicit failure detection
• Examples of “low confidence” vs “high confidence”
• Final demo + explanation

Success = system knows when it might be wrong
40% payment

⸻

Timeline
• Total duration: 2–3 weeks
• Weekly check-ins
• No long-term commitment assumed

⸻

Ideal Candidate

You are a strong fit if you have:
• Shipped ARKit LiDAR / depth systems
• Experience with real-world measurement (even rough)
• Comfort shipping imperfect but honest systems
• Strong communication about limitations

⸻

How to Apply

Please answer:
1. Have you shipped anything using ARKit LiDAR / sceneDepth? Describe or link.
2. At a high level, how would you compute truck-fill bands + confidence from a short walkthrough?

thank you

• Describe a system you’ve built that used depth data (LiDAR / ARKit / ARCore / stereo) to estimate real-world geometry or scale. What failed, and how did you handle uncertainty?
• Given multiple phone captures of a cluttered pile, how would you aggregate depth or geometry to produce coarse volume bands (¼, ½, ¾, full) with a confidence score?

I need someone to fix my rust gstreamer pipeline to make streaming in 1080p60 fps of my live webpage with a video player inside

The proposed system is an intelligent, vision-based threat detection framework designed to operate in real time using surveillance camera inputs. It employs a Spatial–Temporal Graph Convolutional Network (ST-GCN) for multi-person action recognition, enabling the system to analyze human skeletal motion patterns and classify both normal and abnormal behaviors in crowded environments.
To strengthen threat assessment, the action recognition module is complemented by an image classification and object detection pipeline that identifies weapons and potentially dangerous objects, such as firearms or other threatening items, directly from video frames. By fusing behavioral cues with object-level visual evidence, the system achieves more reliable and context-aware threat detection, significantly reducing false positives.
All processing is implemented on a Raspberry Pi–based embedded hardware architecture, integrating a camera module, onboard computation, and peripheral communication units. Once an abnormal action or threatening object is detected, the system flags the individual of interest, tracks their position, and triggers predefined safety responses such as real-time alerts, visual markers, alarms, or secure communication with external monitoring or law-enforcement systems.

Skills: C Programming, Wireless, Engineering, Embedded Software, Machine Learning (ML), C++ Programming, Internet of Things (IoT), Computer Vision, Deep Learning, Object Detection

We are looking for a Machine Learning / Computer Vision specialist to help us design, train, and refine window detection models for real-world environments.

The core challenge is accurate window detection and geometric modulation (position, shape, scale, orientation) from camera input, with the end goal of enabling augmented reality and automated window treatment configuration (e.g. shutters, blinds).

This is not a generic object-detection task. The role requires strong understanding of geometry, scale estimation, perspective correction, and real-world constraints, not just bounding boxes.

You’ll work closely with a product and engineering team to:

Iterate on detection accuracy

Improve robustness across lighting, angles, and window styles

Translate ML outputs into usable geometric data for downstream systems

Key Responsibilities

Design and train computer vision models for window detection

Move beyond simple bounding boxes to precise shape and contour detection

Estimate real-world dimensions from camera input (scale, aspect ratios)

Handle edge cases: reflections, partial occlusion, curtains, frames, arches, angled windows

Improve model performance through dataset refinement and augmentation

Collaborate on iterative model versions (v1 shape detection → v2 semantic understanding)

Provide guidance on model selection vs custom training

Optimize models for real-time or near-real-time inference

Required Skills & Experience
Core Machine Learning / CV

Strong experience in Computer Vision

Hands-on with PyTorch or TensorFlow

Experience with:

Object detection

Semantic / instance segmentation

Contour & edge detection

Familiarity with CNN-based architectures (YOLO, Mask R-CNN, SAM-style models, etc.)

Geometry & Vision Understanding (VERY important)

Understanding of:

Camera perspective & projection

Scale estimation from images

Homography & planar surfaces

Shape normalization and transformations

Ability to extract structured geometric data, not just classifications

Data & Training

Dataset creation, labeling strategies, and augmentation

Evaluating precision/recall for real-world CV use cases

Debugging false positives / false negatives systematically

Bonus / Nice to Have

Experience with AR or spatial computing

Experience detecting architectural elements (windows, doors, facades)

Knowledge of mobile-friendly or edge-optimized models

Familiarity with depth estimation or monocular depth models

What We’re Not Looking For

Pure research profiles with no production experience

Generic “AI generalists” without computer vision depth

Candidates focused only on NLP or LLMs

Engagement Type

Contract / Freelance (with potential long-term collaboration)

Remote friendly

Flexible working hours, outcome-driven

How to Apply (optional section)

When applying, please include:

Examples of computer vision projects

Any work related to geometry, segmentation, or spatial understanding

Brief explanation of how you would approach window detection beyond bounding boxes

We are looking for a skilled developer to create a website extension that utilizes AI technology to transform customer-uploaded images into their dream projects, complete with pricing options. The ideal candidate will have experience in AI development, web technologies, and e-commerce integration. You'll be responsible for building a user-friendly interface that allows customers to upload images and receive instant visualizations and pricing for their projects. If you have a passion for innovative technology and design, we want to hear from you!

• Describe your recent experience with similar projects

We are looking for a reliable Image Annotator to support our computer vision projects by labeling images using polygon annotations on our internal annotation platform.

Responsibilities:
-Annotate images using polygon-based segmentation
-Assign correct annotation classes based on provided guidelines
-Follow visual examples to ensure consistent annotations
-Complete annotation tasks within agreed timelines

Requirements:
-Experience with image annotation or data labeling
-Familiarity with polygon annotationsGood attention to detail
-Ability to follow instructions and guidelines
-Ability to work independently
-Only individual freelancers may apply (no companies or agencies)

Working Hours
Flexible
Approximately 30–40 hours per week

Project Duration"
-Initial test task
-Ongoing collaboration based on performance

Why Join Us?
-Flexible working hours
-Clear task instructions and guidelines
-Potential for long-term collaboration

How to Apply:
Please submit your CV, a brief introduction, and any relevant experience or samples.

• Are you willing to complete a short unpaid test task to demonstrate your skills before we proceed?
• Are you an individual freelancer or an agency? can you commit to 30-40h per week?
• Describe your recent experience with similar projects

We are seeking a reliable, detail-oriented Data Entry & Research Specialist to support our AI data annotation projects. The ideal candidate will be responsible for performing accurate data entry, researching and collecting relevant information, and assisting in annotation tasks for machine learning and computer vision models.

You will work on tasks such as preparing and organizing datasets, supporting labeling and annotation workflows (using tools like CVAT, Label Studio, or similar platforms), and ensuring high-quality and consistent results according to project guidelines. Strong attention to detail, the ability to follow instructions carefully, and a commitment to meeting deadlines are essential.

This role requires someone who is comfortable working with repetitive tasks, has a basic technical understanding of data handling, and can adapt quickly to new tools and task types. Prior experience with data annotation, AI datasets, or research-based data collection is a plus, but training and clear instructions will be provided.

If you are organized, dependable, and interested in working on AI-related projects, this is a great opportunity for long-term collaboration.

Looking for a freelancer to help with a small AI or ML task. This could be a prototype, demo, or micro-model setup. Ideal for testing ideas quickly without building a full production system.

Example tasks:

Running a small ML model (Python/PyTorch/TensorFlow) on sample data.

Testing a pre-trained computer vision model on images or video.

Creating a tiny inference demo for edge or desktop use.

Debugging a small ML/CV pipeline or helping with installation/setup.

Mandatory Skills:

Basic knowledge of Python or C++ for ML/CV.

Familiarity with PyTorch, TensorFlow, or OpenCV.

Ability to run or test pre-trained models.

Understanding of simple ML pipelines (data → model → inference).

Clear communication and following instructions.

Deliverables:

Working demo, script, or snippet showing the model running.

Instructions to run locally or on provided sample data.

Budget: $5 — suitable for small AI/ML prototype tasks or testing models.