https://kapilsh.github.io/Kapil SharmaThis is the website/technical blog of Kapil Sharma. I work at Meta on the Pytorch team. 2026-05-20T10:41:25-04:00 Kapil Sharma https://kapilsh.github.io/ Jekyll © 2026 Kapil Sharma /assets/img/favicons/favicon.ico /assets/img/favicons/favicon-96x96.png Learn CUTLASS the hard way - part 2!2025-12-31T00:00:00-05:00 2026-01-01T08:21:17-05:00 https://kapilsh.github.io/posts/learn-cutlass-the-hard-way-2/ ks Exploring Hopper Architecture on H100 and try to match PyTorch/cuBLAS performance for GEMMs using CUTLASS Learn CUTLASS the hard way!2025-11-01T00:00:00-04:00 2025-11-09T21:47:26-05:00 https://kapilsh.github.io/posts/learn-cutlass-the-hard-way/ ks Walkthrough of optimization techniques for GEMMs from a naive fp32 kernel to CUTLASS bf16 kernel Triton Kernels - Fused Softmax - 22025-09-20T00:00:00-04:00 2025-09-21T07:59:40-04:00 https://kapilsh.github.io/posts/triton-kernels-softmax-2/ ks Worklog: Performance debugging Triton Kernel Triton Kernels - Fused Softmax2025-09-18T00:00:00-04:00 2025-09-24T07:28:40-04:00 https://kapilsh.github.io/posts/triton-kernels-softmax/ ks Fused Softmax Triton kernel exploration Triton Kernels - RMS Norm2025-09-17T00:00:00-04:00 2025-09-18T21:38:06-04:00 https://kapilsh.github.io/posts/triton-kernels-rms-norm/ ks RMS Normalization Triton kernel implementation for LLMs