Hi There.

This is Camel. I engineered a machine learning recommendation engine with TF-IDF vectorization and category-aware similarity scoring, featuring real-time preference learning from user interactions, achieving average accuracy of 92%. I also implemented secure JWT authentication, account personalization, and calendar integration to create the perfect solution to finding campus events - one that can currently serve 10,000+ university students. I automated a data pipeline using Selenium, Apify and Gemini API for scraping and enriching 200+ events daily.

I was working as a contract developer and needed a good way to log my time and know how much I was making at the same time. A MacOS application, it uses local memory to track and store your payouts even after closing out of the app - no more spreadsheets! Low CPU usage: less than the native stopwatch app! Set your own pay rate. Swap between USD and CAD currencies using an API. Great motivator to keep working

Implemented real-time code execution across various languages, saving and viewing code snippets, creating blog posts with code templates, account creation, and commenting and voting on posts, all using a Monorepo strategy. All done through custom API endpoints for each individual task. Then, created responsive and dynamic front-end with dark mode capabilities and an engaging yet simplistic interface.

Used YOLOv5 and PyTorch to train a custom model on a dataset of 120 images, each labeled with one of six common signs from American Sign Language (ASL). The model achieved high accuracy in sign recognition with comprehensive training metrics available.

Web-based literacy assessment platform for grade 1-6 students that utilizes AI in assessing oral production, reading, and writing skills. Built for the Research lab at the Department of Applied Psychology & Human Development in OISE.