Projects n that 🍳

Building cognitive biometric security infrastructure for identity and authentication in enterprise systems. Backed with $100K in pre-seed funding from the 1517 Fund. Converts user inputs into a privacy-preserving 15-D feature vector, mapped to a stable 64-D embedding space via contrastive learning. Cancelable biometric templates using salted hashes.

Fashion app that analyzes your mood boards, camera roll, and daily fit pics to build your style profile. Algorithmically generates context-aware outfits from your existing wardrobe and discovers capsule pieces within your budget to purchase. Product launch hit 150K+ views on Instagram.

Uses AR through Snap Spectacles to bring digital sticky notes into the real world. Tap a location to anchor a note, add a message, and Gemini AI generates a symbolic 3D icon (a book for a study spot, a cup for a cafe). Nearby users see icons anchored in space and can tap to reveal the full note. Built with Snap's World Query Controller, Physics Colliders for finger-overlap detection, and a Pinch Button + TypeScript toggle system for switching between sticky notes and icons.

Gov of Ontario-backed tourist safety startup. Geospatial ETL pipeline ingesting 125K+ unstructured criminology records from global law enforcement portals. Graph Neural Network using PyTorch Geometric to model urban safety as a relational network. Distance-weighted queries across 55+ FBI-standardized crime classifications.

Scholarship and grant finder with a Scrapy-based pipeline aggregating data from 150+ websites. Tinder-style swipe interface as a binary filter for discovery. OAuth 2.0 auto-sync to Google Calendar. Headless browser automation for auto-filling no-essay applications.

Streamlit-powered analytics dashboard tracking Champions League matches with betting insights, live odds movements, and ML predictions.

Looks at apple orchards through DJI drones and tells you what diseases can be spotted. Scans for 30+ pathogen types in apple trees, saving orchardists ~20 hours weekly. Deployed as Android companion.

JARVIS-inspired voice assistant activated by saying "Ronaldo." Uses speech recognition, GPT-3.5, and text-to-speech for natural conversations.

AI music generation based on your mood using Google's SoundStorm architecture. Adjusts frequency, genre, and BPM based on detected mood.

CNN for detecting seven distinct emotions: Happiness, Anger, Disgust, Fear, Sadness, Surprise, and Neutral. Real-time processing with OpenCV. (my first technical project)