Shashwat Sinha

Built a scalable anomaly-detection pipeline for Mac hardware validation, automating failure identification across 150,000+ test runs in collaboration with the AIML team. Engineered a feature-extraction system achieving 99.9% data compression. Implemented detection models (distance-based, Isolation Forests, VAE) with explainable outputs to surface hardware defects for validation engineers.

Architected and prototyped wearable health-monitoring devices end-to-end, selecting biosensors, designing circuits, writing firmware, and building companion software for real-time data acquisition. Coordinated a cross-departmental research effort across three labs, driving hardware and software deliverables from concept through working prototypes; preparing two publications.

Built and deployed an AI-powered voice assistant in Python, integrating speech-to-text and LLM inference for warehouse automation; stress-tested under real warehouse conditions, validating latency and reliability for production use.

Trained and evaluated classification models on wearable biosensor data (heart rate, SpO₂, temperature) for early health-anomaly detection under Dr. Dieff Vital.

Led weekly office hours for ~50 students on MIPS ISA, pipelining, and cache design under Dr. Wenjing Rao.