IshwaryaKavithaRavichandran

GPU systems engineer. I care about the layer where hardware meets model: correctness, reliability, and the infrastructure that makes AI compute trustworthy.

Projects

GPU Correctness Harness ↗

CUDAC++PythonPyTestctypes

CUDA kernels don't throw exceptions when they're wrong. This harness surfaces silent failure modes: warp-shuffle reductions returning half the sum, softmax overflowing to NaN on large logits, tile boundaries that break at N=257. Kernels compiled via nvcc, loaded into Python through ctypes, validated against NumPy CPU baselines.

23/23 tests passing on NVIDIA T4vector_add at 263 GB/s, 82% of peak bandwidth2026

Flash Attention Validator ↗

TritonPythonFP16PyTest

Built and validated a Flash Attention Triton kernel from scratch. Found a non-obvious failure: padding slots in the last K-tile load as 0.0 instead of -inf. exp(0)=1 sits in the softmax alongside real scores. At N=1, that's 63 fake ones against one real score output comes out 64x too small. No crash, no warning, just wrong. Built a harness around that and three other failure modes: tile boundary edge cases, causal mask integrity across block boundaries, and FP16 drift at long sequences.

20/20 tests passing on NVIDIA T42026
Reading

The Power of Now

Eckhart Tolle

On dissolving the noise of the mind and finding stillness in the present moment.

Project Hail Mary

Andy Weir

A lone astronaut, an impossible problem, and the most unlikely friendship in the universe.

Surrounded by Idiots

Thomas Erikson

On why people think and communicate so differently, and how to actually reach them.