How this works▾
Each day is AI-generated by Grok every morning, grounded in the live StartupTribunal codebase, your recent journal entries, and freshly web-searched papers.
At 05:30 Africa/Accra the generator drafts today's card. At 06:00 the day's email lands in your inbox so you can read on your phone before opening the laptop.
Click a day below to open the card. Read the resources, check the code anchors against the live repo, complete the deliverable, take the quiz, write the journal entry. Mark complete. The next day generates against that context.
Static content seeded for days 1-21 as a fallback. AI generation extends this indefinitely along the 4-phase / 30-week arc.
Set your admin key to see your progress overlay.
Today
Why evals exist + install promptfoo
60 min target · Open →
Days completed
0
Current streak
0d
Pct complete
0%
Total days
210
All days
Eval Discipline
0 / 35 doneReplace ad-hoc TDD with systematic LLM evals that catch regressions before they hit prod.
Fine-Tuning & RLHF Intuition
0 / 42 doneGain hands-on training experience so you can discuss fine-tune vs prompt vs router trade-offs without bluffing.
Inference Economics at Scale
0 / 49 doneSpeak fluently on serving cost, latency, and quality trade-offs — vLLM, batching, quantization, KV-cache.
Distributed Systems + Frontier Safety/Interp
0 / 84 doneBuild formalism for enterprise embedding. Read Anthropic's research deeply enough to discuss it as a peer.