Arturo Cortinovis

EdgeTwin — Production IoT monitoring system processing real-time telemetry from 20 simulated factory motors

• Designed end-to-end data pipeline: MQTT telemetry → FairCom Edge ingestion → FastAPI backend → React dashboard with real-time device status, telemetry charts, and alert history.
• Implemented background alert monitoring with hysteresis logic (overheat detection: ≥80°C for 3 consecutive readings; dropout detection: 20-second heartbeat timeout).
• Built custom async HTTP client for FairCom Edge integration; containerized the 4-service architecture with Docker Compose following 12-factor principles.

Express.js + SQLite REST API with full CRUD, validation, and tests

• Designed and implemented a RESTful HTTP API supporting full CRUD with proper HTTP methods and status codes.
• Hardened security: parameterized queries to prevent SQL injection, input validation, global error handler, and 404 fallthrough.
• Wrote 12 endpoint tests (Jest + Supertest) covering happy paths and error cases; added env-based config and clean documentation.

AI agent that analyzes raw footage, detects beats, and assembles cinematic edits

• Building an AI agent that uses vision models to analyze raw video footage, detects musical beats with librosa, and assembles a beat-synced cinematic edit as a DaVinci Resolve project.
• Orchestrates a multi-stage pipeline with ffmpeg: shot analysis, audio beat detection, edit decision logic, and Resolve project export.