Your Skills

• Strong Engineering background with Python development skills with production-quality engineering practices (testing, logging, packaging, code review discipline).
• Proven experience using OpenCV, PyTorch, Torchvision in real projects.
• Experience training models on large datasets for object detection and/or classification, including data balancing, augmentation strategy and performance validation.
• Demonstrated experience shipping computer vision models into production (not only notebooks): integration, monitoring, reliability and iteration cycles.
• Hands-on API/model-serving experience using FastAPI (or equivalent modern framework).
• Ability to design and implement end-to-end pipelines covering data, training, evaluation and deployment.
• Practical experience with fine-tuning workflows and experiment management (hyperparameters, ablations, reproducibility).
• Docker-based packaging and environment reproducibility.
• Basic cloud deployment exposure (any major provider) and comfort with CI/CD workflows.
• Ability to debug performance/reliability issues across the inference stack (pre/post-processing, batching, GPU/CPU utilization, memory, latency regressions).
• Self-driven, motivated and proactive personality

Your Tasks

In this role you will act as a Computer Vision Engineer to build and deploy high-performance vision systems for object detection, OCR and vision-language model (VLM) workflows. You will own model development from data strategy through training, optimization and production deployment with strong engineering standards and measurable impact in real-world use.

• Build production-grade CV systems spanning object detection, classification, OCR and VLM-based extraction/reasoning, ensuring stability, accuracy and predictable latency in real-world conditions.
• Design dataset architecture and scalable pipelines for large training volumes, including labeling strategy, guidelines, QA processes and dataset audits.
• Train and fine-tune models at scale using PyTorch/Torchvision, run systematic experiments, ablations and error analyses to drive measurable improvements.
• Implement MLOps and release discipline: experiment tracking, model registry/versioning, reproducible training, CI/CD, monitoring, rollback plans and incident-friendly operations.
• Adhere evaluation and acceptance criteria by building evaluation harnesses powered by metrics and dashboards for accuracy, robustness, calibration, drift sensitivity, latency/throughput and set release gates for production.
• Deploy models via APIs (e.g., FastAPI) and integrate with backend/product pipelines, storage and downstream workflows to productionize models as services.
• Optimize inference performance using ONNX / ONNX Runtime and TensorRT, applying FP16/INT8 quantization and profiling to meet cost and latency targets.
• Maintain high-quality engineering standards: clean architecture, testing, documentation, performance hygiene and maintainability practices suitable for long-term partnership.

Your Salary

Competitive Salary