Spearheaded development of seven iOS research apps using Swift, Apple Vision Framework, and Google OCR models with on-device CoreML text and face recognition; implemented zero-shot object detection by converting TensorFlow Lite models to CoreML and optimizing for 15 fps across 20 unseen classes; designed IoT connectivity for RayNeo wearable system linking AR glasses with the iPhone app for on-device inference; built a live OCR pipeline and NLP-based search to guide visually impaired users; developed ML-driven bus-stop sign detection and gaze-capture tooling to support clinical workflows; implemented AR/IoT integration to improve reliability and offline capabilities.

IOS App Development Intern at Harvard Medical School, MEEI

August 1, 2025 - September 8, 2025

Spearheaded development of seven iOS research apps using Swift, Zero Shot Object Detection, Apple Vision Framework, and OCR models; integrated CoreML for on-device text and face recognition achieving ~90% accuracy and accelerating study rollout by 25%. Implemented on-device zero-shot detection by converting TF Lite models to CoreML and optimizing inference to 15 fps across 20 unseen classes. Designed an IoT connectivity framework for RayNeo AR glasses, enabling BLE pairing, streaming, and WebSocket feedback to coordinate the iPhone app and AR headset with ~95% directional reliability. Built live OCR pipeline and fuzzy matching for SuperVision Search to highlight queried text with 95% recall. Implemented NLP-based command parsing for navigation, reducing manual input time by ~50%. Delivered ML-driven bus-stop sign detection for AllAboard App with 98% field accuracy and reduced missed stops by 40%. Created 9eye gaze-capture workflow with 30 fps capture and ±2° accuracy, cutting clinici