Inaugural track · now enrolling
Ship the Stack.
Build a working RAG system over your own documents. Six modules. Six ship gates. The benchmarks are the scoreboard.
How the track works
- Upload your corpus.
- Read the module. Build the thing.
- Your ship gate runs. It passes or it does not.
- Beat the boss challenge. Unlock the next module.
Other tracks
More tracks land after the RAG & Evals cohort closes. Same pattern: build the thing to learn the thing.
- Planned Agents & Tool Use Build a production-grade agent loop with tool retries, cost caps, and observable traces.
- Planned Evals at Scale Ship a live eval harness that gates deploys. Judge prompts, regression sets, human-in-the-loop.
- Planned Prompt Engineering Ops Prompt versioning, caching, routing, cost attribution. The unglamorous half of production AI.