Life-Dashboard
Sep 2025 – Dec 2025
PythonFastAPIPostgreSQLReactTypeScriptVertex AI Gemini
Highlights
- Designed an agentic nutrition workflow that decomposes natural language into recipes + ingredients, uses web search to normalize/cache nutrition facts, and tracks intake vs personalized goals (with 95% CIs).
- Preprocessed 14-day time series of biometric data from the Garmin API to drive AI-curated recovery insights (sleep, HRV, RHR, training load).
- FastAPI backend ingests Garmin HRV, resting HR, sleep, and training load into PostgreSQL using async SQLAlchemy + Alembic migrations.
- React/TypeScript dashboard surfaces training trends, nutrition workflows, and intelligent to-dos with scheduled Garmin refresh flows.