Development runs in milestones — each a coherent, shippable slice. Commits and PRs reference their milestone (e.g. "Part of M2"), and progress is tracked in GitHub Milestones and Issues.
| Milestone | Scope | Status |
|---|---|---|
| M0 — Repo & scaffolding | License, CI, CONTRIBUTING, CODEOWNERS, templates, monorepo layout |
✅ Done |
| M1 — Training pipeline (Python) | Dataset build, taste training, prediction core, ONNX export, demo workbench | ✅ Done |
| M2 — .NET API + ONNX serving | ASP.NET Core skeleton, load taste ONNX, /predict, in-process ONNX Runtime |
🔜 Next |
| M3 — React workbench | React UI against the fixed JSON contract; taste meters with confidence tags | Planned |
| M4 — Aroma model | Commercial: aroma trains on licensed (PMP 2001) / customer data. The full open aroma model moves to the academic edition — deferred here, see docs/AROMA.md | Deferred → academic |
| M5 — Packaging | Dockerfiles, docker-compose, Postgres + pgvector, single-box deploy |
Planned |
| M6 — Pilot-ready | Auth + per-seat, pgvector substitution search, polish, demo script | Planned |
See docs/ARCHITECTURE.md for the full rationale.
- Track A — Python training + demo. Largely complete (M1): the model pipeline plus a lightweight FastAPI/HTML workbench.
- Track B — the .NET/React product. M2–M6: Python trains → exports ONNX → ASP.NET Core serves in-process → React workbench → Postgres/pgvector → Docker Compose on a single box.
Both tracks consume the same trained model artifacts.