Achieving 96%+ accuracy on US state classification with a 7-model ensemble strategy
NYU Computer Vision project using Vision Transformers, CLIP, and ensemble techniques to predict geographic locations from street view images.
NYU Machine Learning project comparing scaling laws of Transformers vs LSTMs on symbolic music. Trained models from 1M to 100M parameters on Irish folk tunes.
NYU Computer Vision project using template matching and image processing to detect and identify star constellations in astronomical images. Demonstrates classical CV techniques and their applications.
A deep dive into implementing real-time people detection and tracking using YOLOv8/v10 and DeepSORT, with deployment on Azure cloud platform. This project demonstrates end-to-end ML engineering from model selection to production deployment.
Semi-supervised learning is crucial when labeled data is scarce. This post explores my implementation of MixMatch, a state-of-the-art semi-supervised learning algorithm that combines consistency regularization with pseudo-labeling.
A deep dive into implementing real-time people detection and tracking using YOLOv8/v10 and DeepSORT, with deployment on Azure cloud platform. This project demonstrates end-to-end ML engineering from model selection to production deployment.