Algorithmic Mutterings

Writings on CS, math, and life.

My Solutions to "How to scale your model"

My Solutions to "How to scale your model"

11 min read · July 15, 2025

2025 · comments · interview llm inference scale
30 Interview Questions on LLM Inference

How many FLOPs is attention? When does a KV cache hurt performance? Why does beam search cause your KV cache memory to explode?

28 min read · June 30, 2025

2025 · comments · interview llm inference kv-cache
Can You Solve Dynamic Programming Problems Faster? Part 1, One-Dimensional LWS

We prove you can solve dynamic programming problems polynomially faster if you have a simple cost function for kD LWS problems.

22 min read · January 9, 2024

2024 · comments · explain-paper dynamic-programming cs-theory algorithms complexity