Selected work / Case study
ChatGenius
Slack-like chat with AI replies.
Frontend tests checkedOverview
A chat application built with React, Supabase, FastAPI, and pgvector. Code only; no public demo is linked.
- React
- Supabase
- FastAPI
- pgvector
The problem
ChatGenius explores Slack-like communication with channels, direct messages, attachments, and AI replies based on message history. It combines the chat interface and AI response paths in one codebase.
My contribution
I built a React chat frontend and a Python API for AI replies and related chat operations. The source includes channel and direct-message organization, presence, threads, and file sharing. This portfolio presents the code only, without a live demo.
How it's built
The application lives in frontend/ and backend/; the root React starter is separate. The frontend uses Supabase for authentication, data, and live updates. FastAPI chat routes retrieve Supabase message history and call OpenAI directly. A separate SQLAlchemy package implements vector retrieval with PostgreSQL and pgvector.
What's verified
Frontend: 55 of 55 tests pass.
Eyrie reported all 55 frontend tests passing on October 29, 2026. Backend services and a live chat flow are not verified here. Code only; no demo link.
What remains unverified
The passing frontend tests do not establish full type correctness or a working chat service. Standalone type checking fails, backend tests fail before assertions, and database migrations need reconciliation. Authenticated messaging, uploads, and AI replies remain unverified.
Captured screens
