12-Month On-Ramp · July 2026 → July 2027

Ship AI you can defend.

"I don't just call the model. I measure it, retrieve for it, watch it in production, and keep it safe — and I can prove every line."

An interactive, roadmap.sh-style plan that turns your existing repos into the vehicle for the four disciplines that separate people who use LLMs from people companies hire to work on LLMs: evals, retrieval, observability, and safety — plus the mini-project catalog (Tracks A & B) and the client-evidence engine added by the July-2026 verification and adversarial-review passes. Click any node to mark it done — progress saves to this browser.

0%
Overall progress
0 / 0 nodes complete · your journey is saved locally
Core (non-negotiable) Recommended Optional / stretch Portfolio artifact Explain-it-cold drill