kushaalveijay.dev
Grokking AI System Design16 lessons2026-10-01
- track index →
- 01Tokens: The Unit of Everything13 min
- 02Embeddings: Meaning as Geometry7 min
- 03Attention: Why Context Is Expensive13 min
- 04The Generation Loop: Prefill and Decode7 min
- 05The KV Cache: The Memory That Runs the Bill8 min
- 06The Context Budget8 min
- 07The Retrieval Gap11 min
- 08Chunking: The Irreversible Decision9 min
- 09Two-Stage Retrieval: Recall Then Precision8 min
- 10The Agent Loop8 min
- 11Tools as Contracts12 min
- 12Memory: What to Keep8 min
- 13Evaluation: The Eval Is the Product9 min
- 14Failure Is the Default8 min
- 15The Cost Model9 min
- 16Putting It Together18 min