Roadmap
This is a single-tenant portfolio project; some features only make sense in a multi-user product and live in Phase 2 candidates below, called out honestly.
Shipped in v1
- Six ingestion sources end-to-end with idempotent upserts
- Bronze/silver/gold Postgres warehouse on Supabase
- dbt models with not_null/unique/accepted_values tests on every grain
- SCD2 dim_assessments via dbt snapshot
- Prefect 3 flows on cron schedules in Australia/Adelaide
- Streamlit dashboard with five pages and password gate
- Anthropic-backed summariser, flashcards, citation suggester with DOI validation
- Daily Resend digest and weekly summary
- Synthetic data generator so demos work without credentials
- Both subdomains deployed independently; CI workflows separated
Phase 2 candidates (not in v1)
- Citation graph visualisation on the Readings page
- PDF full-text extraction from Zotero attachments
- Discord notifications wired up (currently stubbed, disabled)
- OpenAI fallback for AI features
- AnkiConnect integration (currently TSV export only)
- Mobile-first responsive polish on the Streamlit app
- Multi-language note support
- Spaced-repetition scheduling beyond raw flashcard export
- Browser extension for one-click reading capture