Chester Corin

Automating sensitive personal healthcare workflows without compromising data privacy through external cloud-based AI providers.

Architected a fully on-premise AI cluster. The decentralized setup leverages distributed inference across local hardware to securely process private health data and orchestrate intelligent automated workflows, completely removing cloud dependency. Currently experimenting with various agentic platforms to further enhance the automation capabilities, and personalized assistance.

Football outcomes are inherently difficult to forecast. Form fluctuates, context matters, and relying on a single signal — whether statistical or market-based — leaves blind spots that reduce prediction accuracy.

A Premier League prediction engine that blends three independent probability models (XGBoost, ELO Ratings, and Market Consensus) into calibrated forecasts. Every prediction is recorded in a versioned ledger so accuracy and calibration can be tracked transparently over time.

Raw broadcast video is unstructured and time-consuming to analyze manually. Coaches and analysts spend hours scrubbing through footage, tagging players, and estimating positions.

An automated pipeline that ingests standard broadcast video, detects and tracks every player and the ball, classifies players into teams, and renders annotated output with trajectories and stats overlays.

Combating the overwhelming urge to "do it all" at once, which often leads to burnout and a loss of focus on the daily process.

A privacy-first iOS application that acts as an intelligent life coach. By utilizing Apple's local Foundational Models (CoreML), it offers personalized guidance to help users find their rhythm, ensuring all sensitive reflections and data remain strictly on-device. The project is currently serving as an active testing ground for advanced agentic AI coding workflows.

Evolving the passive "rubber duck debugging" technique into an interactive experience that physically resides on the desktop.

A 3D-printed, Raspberry Pi-powered companion that has grown from a simple weather widget into an LLM-integrated assistant. Flo listens to spoken problem descriptions and replies with randomized or context-aware questions, making the debugging process a true conversation.