Summary

Turns Streamify from an Eventsim/GCP-only demo into a local-first Yandex Music self-analytics product while preserving the legacy data engineering track.

Tracks #4
Tracks #5
Tracks #6
Tracks #7
Tracks #8

Product Value

• Ingests personal Yandex Music metadata locally without downloading or storing audio.
• Builds DuckDB/dbt marts for favorite artists/tracks, genre shifts, playlist overlap, repeat signals, diversity, active periods, underrated tracks/playlists, and data quality.
• Adds Streamlit dashboard plus markdown, JSON snapshot, and CSV action queues.

Data Engineering Impact

• Raw/Bronze: token-safe metadata ingestion, JSONL/Parquet outputs, manifest row counts and SHA256 checksums.
• Silver/dbt: typed local staging with robust manifest diagnostics parsing.
• Gold/dashboard: local marts, Streamlit app, static exports, recommendation queues.
• Privacy/security: .env/raw/DuckDB/audio guards, sample-only CI/Pages/release artifacts.

GitHub Delivery

• Adds issue templates, PR template, project management docs, release process, release notes.
• Adds GitHub Pages workflow and tag-based release workflow.
• Enables issues and creates agent-lane tracking issues [Agent][Repo/Build] Stabilize local product build and delivery automation #4-[Agent][QA/Integration] Gate real-account readiness and release safety #8.

Checks

• make test
• make acceptance-real locally with private .env token
• make pages-site
• GitHub Actions validate-local-product-contract
• Real raw data, DuckDB, reports, recommendations, public output and .env remain ignored

Notes

Yandex Music live ingestion uses an unofficial client library. Public automation intentionally runs on sample metadata only.