Problem

I wanted a realistic financial data product that combined backend services, a typed frontend, scheduled ingestion, and AI-assisted data entry. The challenge was to make stock and portfolio workflows feel operational instead of just displaying a static table.

Approach

• Built a Go REST API with migrations and background ingestion for stock records, quotes, recommendations, and watchlists.
• Used CockroachDB locally as the SQL store, managed by Docker Compose for repeatable setup.
• Added a Vue 3 + Vite + TypeScript frontend with Pinia state and Tailwind styling.
• Integrated Gemini to extract portfolio positions from brokerage screenshots.
• Added scheduled fundamentals and quote refresh flows using Financial Modeling Prep, Alpha Vantage, and FRED-style yield inputs.
• Split fundamentals work into a small FastAPI service so Python data tooling could live beside the Go API without turning the backend into a script pile.

Results

• End-to-end containerized stack that can be brought up locally with one docker compose up --build.
• Portfolio upload flow turns an image into structured positions for downstream portfolio views.
• Recommendation endpoints combine cached quotes, fundamentals, and valuation inputs instead of relying on one raw provider response.

Links

• Source repository