Blog

Long-form notes on systems, languages, and the things I build. Multi-part series live under their own page.

Scaling from 100 to 100 Million Users

What I learned tracing what actually breaks as a system grows: how to tell which part is struggling, the smallest fix that works, and what every fix quietly costs.

System Design
Scalability
Distributed Systems

Kaushik SamadderOct 15, 202669 min read

Series · How to Make an LLM From Scratch to Inference

Part 1: Building a Byte-Pair Tokenizer

Text goes in, integers come out. Building BPE from scratch and understanding why vocabulary size is a real tradeoff.

LLM
Tokenization

Kaushik SamadderJul 24, 20261 min read

Series · How to Make an LLM From Scratch to Inference

Part 2: Attention, Implemented Twice

The slow, obvious version first — then the batched one, and a look at what actually changed.

LLM
Attention

Kaushik SamadderJul 24, 20261 min read