At CamSentra, I architect and deploy production-grade computer vision platforms that process real-time multi-camera RTSP video for security, retail, and industrial enterprises.
I've built end-to-end MLOps pipelines spanning CCTV data collection, Roboflow annotation, YOLO and CNN-LSTM model training on GCP/Colab GPUs, and dataset versioning in Google Cloud Storage. I optimize edge performance through CUDA configuration, Docker containers, and thread-isolated RTSP decoding for sub-second latency.
My work includes NeuraSense, apartment security, retail theft analytics, and industrial PPE and restricted-area monitoring, with FastAPI/Flask services connected to React dashboards and Flutter mobile apps.
