Data + ML

Concepts, Notes and Intuitions

GPU Essentials - A Concise Technical Guide

A concise, technical guide to GPU architecture and CUDA, showing how massive parallelism is achieved through threads, blocks, SMs, and memory hierarchies.

26 min read · 2025

Wrapping Up Our ML Foundations Journey

A reflection on our exploration of machine learning fundamentals, from mathematical prerequisites to gradient boosting.

3 min read · 2025

CuTe DSL fundamentals and primitives [FA3]

The CuTe Python DSL and Hopper primitives behind the FlashAttention-3 worklog, sourced from the PTX ISA and the CUTLASS source, from layouts and tensors to TMA, mbarriers, pipelines, WGMMA, TV layouts, the register handoff, warp specialization, pipeline ordering and FP8

87 min read · June 07, 2026

2026 · GPU · GPU-NYU
Graduated from NYU with MSCS

Reflections on 2024-2026 Master's journey.

3 min read · May 22, 2026

2026 · Life
FlashAttention 3 - A Worklog

Implementing FlashAttention 3 from scratch

119 min read · May 15, 2026

2026 · GPU · GPU-NYU
CUTLASS WGMMA on Hopper - Notes

My notes on WGMMA internals from the Colfax Research CUTLASS Hopper GEMM blog

112 min read · April 18, 2026

2026 · GPU · GPU-NYU
Investigating Flaky `test_eagle_dp` — Batch Invariance Failure on L4 GPUs

Investigative notes and fixes for test_eagle_dp CI tests in vLLM

40 min read · April 14, 2026

2026 · GPU · GPU-NYU

Data + ML

Concepts, Notes and Intuitions

GPU Essentials - A Concise Technical Guide

Wrapping Up Our ML Foundations Journey

CuTe DSL fundamentals and primitives [FA3]

Graduated from NYU with MSCS

FlashAttention 3 - A Worklog

CUTLASS WGMMA on Hopper - Notes

Investigating Flaky `test_eagle_dp` — Batch Invariance Failure on L4 GPUs