Posts

2025  11

July  2

CUDA really isn’t that bad: Tiling, Fusion, and Triton

July 11, 2025 · 22 min · ML Theory + Code

CUDA really isn’t that bad: Kernel Ops and Memory Hierarchy

July 3, 2025 · 20 min · ML Theory + Code

June  3

Designing & Building an Agentic AI framework from scratch

June 30, 2025 · 20 min · ML Systems + Code

Tesla’s Robotaxi > Google’s Waymo: Vision vs. LiDAR

June 26, 2025 · 10 min · Computer Vision & Personal Opinions

Deploying a toy ML model to production

June 14, 2025 · 11 min · ML Systems + Code

May  2

Fun with LoRA: How low-rank can we go before adjacency matrices break down?

May 27, 2025 · 21 min · ML Theory + Code

Designing AlphaTicTacToe (and AlphaTicTacToeZero)

May 2, 2025 · 20 min · ML Theory

April  1

Can a EMNIST model run on an Amazon Kindle from 2012?

April 10, 2025 · 13 min · ML Systems + Code

March  1

Building Transformers from scratch: Multi-Head Attention, LayerNorm, and the brain behind ChatGPT

March 13, 2025 · 19 min · ML Theory + Code

February  1

How did we get to Transformers? The rise of Attention

February 28, 2025 · 18 min · ML Theory

January  1

Word2Vec from scratch: Intuition to Implementation

January 10, 2025 · 14 min · ML Theory + Code

2024  1

December  1

My goals for this blog

December 29, 2024 · 1 min · Personal thoughts