Appearance
90-Day Learning Journey
Thirteen weeks · one continuous loop
A deliberate progression from model fundamentals to a production-minded AI system. Each week produces an artifact, a measurement, and a reflection.
Learn↓Understand↓Design↓Build↓Evaluate↓Reflect
03progressive phases
13focused weeks
01integrated system
Phase 1 — Foundations
Days 1–28 · Build the vocabulary
Understand model behavior, prompts as interfaces, structured generation, embeddings, retrieval, and the basics of measurement.
Phase 2 — Systems
Days 29–63 · Connect the components
Design retrieval pipelines, tool contracts, agent loops, trace schemas, and repeatable evaluation suites.
Phase 3 — Production
Days 64–90 · Engineer for reality
Add reliability controls, safety boundaries, cost discipline, release gates, and operational feedback.
Weekly operating rhythm
| Day | Focus | Evidence of progress |
|---|---|---|
| 1 | Learn | A concise concept map |
| 2 | Understand | A mechanism explained from memory |
| 3 | Design | A decision record with alternatives |
| 4–5 | Build | A working vertical slice |
| 6 | Evaluate | Results and error categories |
| 7 | Reflect | One durable lesson and next change |
Definition of done
A week is complete when the artifact runs, its behavior is measured, one failure is understood, and the next iteration is written down. Time spent is an input; evidence is the output.