Project 01 / AI Systems
A personal AI operating system that runs on real infrastructure and does real work — not a demo, not a prototype, a system I actually use.
Jarvis is my personal AI assistant running on hybrid cloud and local infrastructure. The goal: zero manual effort for detecting, scheduling, and completing tasks.
It scrapes Google Classroom automatically, detects new assignments via OCR from my iPhone, generates draft responses using an LLM, and schedules work based on my energy levels throughout the day.
Everything runs on GCP Cloud Run with a PostgreSQL database on Cloud SQL, triggered by Cloud Scheduler every minute.
FastAPI backend deployed on Cloud Run. Cloud Scheduler fires every minute hitting /run-checks/execute. Cloud SQL Postgres connected with Secret Manager credentials.
Working on: jarvis-sweep Cloud Run Job to connect the scheduler to real execution — Playwright scraper, LLM draft generation, and frontend endpoints.
Estimated MVP gap: 10–18 hours of focused build time.