Hire the Best OpenCV Developers
in the United Kingdom

I’m Niloy, an AI engineer specialising in computer vision, deep learning, generative AI and full-stack ML systems. I build high-performance, production-ready AI solutions for companies that need reliable results, not experiments. I work across the full lifecycle: scoping requirements, architecting systems, preparing datasets, training and fine-tuning models, and deploying to scalable cloud infrastructure. My recent work includes building multi-agent systems, HIPAA-compliant AI chatbots, end-to-end computer vision platforms, OCR pipelines for real-world documents, and fully automated systems. 🔧 What I Do • Develop and deploy deep-learning systems: classification, detection, segmentation, retrieval, embeddings, video analytics • Implement RAG systems, custom tools, and hybrid search architectures • Build voice/telephony AI agents for customer service, intake, and automation • Build and fine-tune LLMs, VLMs, multimodal pipelines, and agent-based AI systems • Train OCR models for real-world documents, ID cards, license plates, handwritten text and noisy images • Create synthetic data pipelines and data-augmentation workflows • Design scalable ML backends with FastAPI, AWS/GCP, Docker, CI/CD • Integrate lightweight robotics/embedded CV pipelines when needed 🧰 Technical Stack • Python, PyTorch, TensorFlow, Sklearn, HuggingFace, OpenCV, LangChain/LangGraph, Pipecat • FastAPI, Docker, Kubernetes, GitHub Actions, AWS, GCP, DigitalOcean, Slurm • Postgres, SQL, Pandas, NumPy • Experience with Jetson, Raspberry Pi, ROS2 (for vision-edge use cases) ⭐Selected Work Highlights • Architected a HIPAA-compliant multi-agent system for insurance claims processing using FastAPI, LangGraph and AWS Bedrock • Built vision-based self-checkout (VisionPOS) for low-resource pharmacies; reached 95 percent accuracy in real pilots • Developed multiple production OCR pipelines, including license-plate recognition, ID document extraction, handwritten prescription OCR, and satellite-image super-resolution • Delivered AI voice agents for sales, intake, data entry and customer service integrated with telephony systems • Built reverse image/video search systems and large-scale embedding pipelines • Trained deep learning models for medical imaging and clinical decision support ⚙️How I Work • Clear requirements and structured communication • Reproducible, well-documented code with version control • Milestone-driven delivery • Production mindset: reliability, latency, safety, and maintainability first