
Projects
A curated collection of systems I’ve built, from real-time trading engines to mobile-first AI apps.

ChewVue
SwiftUI |CoreML |iOS |ChewML
AI-powered food recognition and calorie tracking, running 3× faster than cloud alternatives via on-device CoreML

AI-powered food recognition. Machine learning engine behind ChewVue, enabling on-device food recognition and real-time nutritional insights. Features a custom CNN, transfer learning, and automated pipelines for continuous model improvement and deployment via CoreML. Calorie tracking, runs 3× faster than cloud alternatives via on-device CoreML

Vocabulary-building app with word games, user profiles, favorites, and daily challenges. Built natively with UIKit.


AlgoStream
OCaml |Monte Carlo |Time Series
Statistical arbitrage engine with under-5ms execution latency and advanced backtesting tools.




Scalable C++ neural network framework with support for distributed training, federated learning, multi-modal data, and cloud deployment. Built using Eigen for high-performance matrix operations. Designed with a modular architecture, SGD and Adam optimizers, and both classification and regression loss functions. Features model saving/loading, MPI-based distribution, OpenCV integration, and AWS support.

Deep learning image classifier for identifying flower species. Trained using VGG16/ResNet architectures and optimized processing pipelines for mobile inference. Designed with command-line interface and deployment-ready architecture.


Tan4
Google Gemini |GCP |Deepfake Detection
Advanced AI-powered deepfake detection system built for fintech. Tan4 safeguards digital identity verification by detecting synthetic audio and video in real-time using Google Gemini’s ML infrastructure. Designed for account onboarding, fraud prevention, and compliance in high-risk financial environments.