Planning

Planning

Planning is managed through GitLab issues, milestones, merge requests, and sprint reviews. The durable planning model is:

1. Keep the project vision and success criteria in the Charter.
2. Break product work into small issues tied to workflow phases or technical systems.
3. Use merge requests for implementation review and CI evidence.
4. Update wiki/product documentation when behavior, setup, APIs, or architecture changes.
5. Use benchmarks, evals, screenshots, and reports as project evidence rather than relying only on verbal demo notes.

Primary planning themes:

• end-to-end ML workflow completeness;
• human-in-the-loop LLM orchestration;
• data/document ingestion and search;
• notebook-backed execution;
• model evaluation and deployment;
• developer quality gates and reproducible demos.