I have hands-on experience in building and deploying real-time Computer Vision pipelines using OpenCV, NVIDIA DeepStream, TensorRT, and Docker.
I have worked with Ultralytics-based Docker environments, explored their deployment workflows, and contributed to open-source projects through debugging, testing, and improving existing pipelines. I also have experience creating my own custom Docker images for GPU-based inference setups, ensuring reproducibility across systems with CUDA, TensorRT, and OpenCV dependencies. My work includes optimizing deep learning models using FP16/INT8 acceleration, converting models into TensorRT engines, and integrating them into DeepStream/GStreamer multi-stream video analytics pipelines.
Along with this, I am comfortable with Linux-based development, performance benchmarking, and edge deployment. I am highly motivated to contribute to OpenCV GSoC, and I am ready to work on the idea: Expanding kornia-vlm with ONNX and TensorRT backends for efficient, production-ready inference.