Abhik Sarkar

Director of Machine Learning at Cloudastructure. Computer vision for physical security; GPU performance and the unglamorous parts of ML systems.

I write about CUDA, NVIDIA driver internals, and numerical stability at abhik.ai.

Recent writing

• Best Resources for Learning CUDA Matrix Multiplication Optimization _{Jun 03, 2026}
• C++ Build Pipeline: Compilation vs Linking vs Loading Explained _{Jun 03, 2026}
• H.264 vs H.265 vs AV1: Comparing Modern Video Codecs _{Jun 03, 2026}
• CUDA Matrix Multiplication Optimization: From Naive to Near-cuBLAS _{Apr 07, 2026}
• The Complete NVIDIA Xid Error Field Guide _{Mar 13, 2026}

More on abhik.ai →

Currently

• Building pylings.
• Deeper on CUDA kernel authoring and Nsight Systems workflows.
• Researching failure modes in ML training clusters (GPU, network, and storage); wrote the NVIDIA Xid Error Field Guide from that work.
• Upcoming: talk at EuroPython 2026 (July, Kraków).
• Workshop at PyCon Italia 2026: "Write Your First High-Performance GPU Kernel in Python!" (github).
• Workshop at PyCon India 2025: ArrPy: rebuilding NumPy from scratch.

Elsewhere

abhik.ai · X · LinkedIn