Computer Vision Engineer - Real-Time Video Analytics

Location: Remote  

Type: Contract / Full-Time  

Experience : 2+ years

-----

## About the Role

We’re building a production-grade real-time video analytics platform that processes multiple camera feeds with GPU-accelerated inference. You’ll work on a microservices architecture handling high-throughput video processing and multi-stage detection pipelines.

-----

## Technical Stack

### Core Technologies

- *Python 3.10+* - Primary development language

- *FastAPI* - Async API services and WebSocket

- *Docker & Docker Compose* - Containerization and orchestration

- *Apache Kafka* - Message streaming and event processing

- *MinIO* - S3-compatible object storage

- *Redis* - Caching and real-time data

- *PostgreSQL* - Analytics and time-series data

### Computer Vision & ML

- *Object Detection Models* - YOLO (v8/v11), RT-DETR, Faster R-CNN, or similar architectures

- *NVIDIA Triton Inference Server* - Production inference serving

- *TensorRT* - GPU-optimized model deployment

- *Object Tracking* - ByteTrack, DeepSORT, BoT-SORT, or custom implementations

- *OpenCV / CV2* - Image processing and manipulation

- *Model Training & Fine-tuning* - PyTorch, TensorFlow, or other frameworks

### Video Processing

- *FFmpeg* - Video decoding with NVDEC hardware acceleration

- *GStreamer* (optional) - Advanced video pipeline processing

- *RTSP protocols* - Live camera stream handling

### Monitoring & DevOps

- *Prometheus & Grafana* - Metrics and dashboards

- *NVIDIA DCGM* - GPU monitoring

- *Docker deployment* - Production containerized services

### Frontend (Nice to Have)

- *React 18* - Dashboard development

- *WebSocket* - Real-time client updates

-----

## What You’ll Work On

- Build and optimize GPU-accelerated video processing pipelines

- Implement microservices for real-time inference and event processing

- Deploy and configure NVIDIA Triton for batched inference

- Evaluate, train, and export state-of-the-art detection models to TensorRT

- Design Kafka-based event streaming architecture

- Implement and optimize multi-object tracking algorithms

- Develop monitoring and observability solutions

- Write production-ready, scalable Python services

- Research and integrate new computer vision techniques as they emerge

-----

## Required Skills

*Must Have:*

- Strong Python development (async/await, multiprocessing)

- Experience with modern object detection models (YOLO, RT-DETR, R-CNN family, or equivalent)

- Deep learning model deployment (TensorRT, ONNX, OpenVINO, or similar)

- Docker containerization and GPU workloads

- Message queue systems (Kafka, RabbitMQ, or similar)

- Linux environment and command-line proficiency

*Strongly Preferred:*

- NVIDIA Triton Inference Server experience

- Real-time video processing with FFmpeg or GStreamer

- Object detection and tracking implementations

- Microservices architecture patterns

- Prometheus monitoring and metrics

- Production ML systems at scale

*Nice to Have:*

- FastAPI or similar async Python frameworks

- React or modern frontend experience

- PostgreSQL and time-series databases

- Kubernetes or container orchestration

- Edge computing deployments

-----

## Qualifications

- Bachelor’s degree in Computer Science, Engineering, or related field (or equivalent experience)

- 2+ years of professional software development

- 1+ years working with computer vision or ML systems

- Portfolio or GitHub showing relevant projects

- Strong problem-solving and debugging skills

- Excellent communication for remote collaboration

-----

## Project Context

You’ll join a project that’s already architected with:

- Two-stage detection pipeline design

- Microservices ready for horizontal scaling

- Production monitoring from day 1

- Clear path from demo (2 cameras) to production (100+ cameras)

Initial focus: Build core services, optimize GPU utilization, ensure system reliability.

-----

## Why Join?

- Work on cutting-edge computer vision technology

- Production-scale GPU inference challenges

- Modern tech stack (latest YOLO, Triton, TensorRT)

- Real impact on deployed systems

- Collaborative environment with clear architecture

-----

## To Apply

Send to: opash.hr@gmail.com

Include:

1. Resume / CV

1. GitHub profile or code samples

1. Brief description of your most relevant computer vision project

1. Availability and rate/salary expectations

*Subject Line:* “Computer Vision Engineer - [Your Name]”

-----

## Interview Process

1. *Technical Screening* (30 min) - Discuss experience with stack

1. *Coding Challenge* (take-home) - Small video processing task

1. *Technical Deep Dive* (60 min) - System design and architecture

1. *Final Discussion* (30 min) - Project details and logistics

-----

• *We’re looking for someone who can start contributing within 1-2 weeks of onboarding.*