I am Abhik Sarkar, a machine learning engineer focused on building real-world computer vision systems that actually run at scale. My work lives at the intersection of software engineering, GPU hardware, and production reliability .
I currently lead machine learning at Cloudastructure, where I design end-to-end vision pipelines spanning high-throughput video ingestion, GPU-accelerated decoding, and low-latency inference. My daily toolset includes PyTorch, TensorRT, NumPy, OpenCV, CuPy, PyCUDA, and ONNX Runtime.