
bastaFDA
[shipped]2025 – 2026Capstone Project — defended before a faculty panel
A mobile app that verifies medicine authenticity by cross-referencing OCR text extraction and on-device image recognition against official FDA registration data.
- Trained and integrated a TensorFlow Lite image-classification model to identify supported medicine products on-device.
- Engineered a multi-scan OCR workflow that aggregates results across captures, improving accuracy on low-quality packaging.
- Implemented the full product flow — scanning, verification, history, reporting — with Firebase authentication and real-time data.
- Leveraged agentic coding tools (Claude Code, GitHub Copilot) for scaffolding, refactoring, and test iteration, reviewing and validating all generated code against project requirements.
