Akshay's Expedition Logs

Deep dives into distributed systems, ML infrastructure, and the architectures that make AI at scale actually work

Featured

14 min read•2026-04-18

Attention is a Potluck

A tactile, phone-readable walkthrough of the single mechanism that makes every modern LLM actually understand language — no matrices first, no jargon, no mush.

Why Your AI Images Look Pixelated — Quantization, Distillation & Inference

GenAI

14 min read•2026-03-18

Why Your AI Images Look Pixelated — Quantization, Distillation & Inference

I spent 18 minutes generating an image that looked like a bad oil painting. Here's the exact threshold where quantization breaks down — and why distilled models can't use negative prompts.

Production Ray: Patterns, Mistakes, and Lessons

Distributed Systems

•2026-03-15

Production Ray: Patterns, Mistakes, and Lessons

The final part of our Ray deep dive: five production patterns that work, five mistakes that hurt, and a debugging playbook for distributed ML systems.

Running AI Image Generation Locally on Apple Silicon

GenAI

•2026-03-08

Running AI Image Generation Locally on Apple Silicon