Laya & Jev.
Decide before generating.
Understand the decision engines, run Laya in Portuguese, and build a triage you can evaluate.
From question to verifiable decision.
Read in order or jump straight to the track that resolves your question. Set aside about 5–7 hours with the labs; the time varies with your experience and the installation.
🧭 Fundamentals
Understand what the decision solves before installing any model.
Explore track → TRACK 02 · 2 MODULES⚙️ Local engine
Move from a design on paper to real inference in Portuguese.
Explore track → TRACK 03 · 2 MODULES🛠️ Practical application
Turn predictions into a verifiable triage console.
Explore track → TRACK 04 · 2 MODULES📊 Quality and evolution
Measure errors, calibrate expectations, and plan specialization.
Explore track →What you will build
A local support desk with interface, API, terminal execution, and evaluation in labeled tickets. The model suggests; the policy keeps human review.
For the practical part: terminal, Python, and basic JSON knowledge. The conceptual lessons start from zero.
Evidence, not universal score.
In the local evaluation, the model got 13 out of 16 synthetic tickets correct. Three mistakes were preserved for analysis. The result confirms real execution and shows why review is still necessary.
Read sources, caveats, and evaluation →Read your way
Mark topics as read, flag questions, and select excerpts to highlight or annotate. My learning journey brings everything together and lets you export/import JSON. The state stays in this browser; make a backup before switching devices.