The AI Infra Dev study hall

Learn the systems.
Understand the tradeoffs.

Go deeper into the engineering behind AI. Courses built around first principles, working code, and experiments you can explain.

Explore the courses

The course collection

01 course · growing over time
Course 01 In development

Performance engineering for LLM inference

Build an inference engine. Predict its bottlenecks. Measure what happens, from a single kernel to a distributed serving system.

Performance modelingGPU kernelsInference systems
Learning path
8 projects
Suggested pace
16 weeks
Level
Advanced
Qwen3 · 8B & 32BFrom kernels to systems

A living collection. The first course is being developed in the open; more subjects will follow.

What we’re building toward

More than a reading list.
A place to work things out.

Draft materials are available now. The online study experience will grow alongside the courses.

Planned

Read with context

Online chapters that bring explanations, code, and diagrams together.

Planned

Learn by experimenting

Guided labs to test a hypothesis and explore what changes the result.

Planned

Make the learning yours

A place for study notes, checkpoints, and the questions you want to revisit.