• – Engineered data and validation pipelines using Python, Pandas for feature discovery and edge-case testing for LightGBM-based cross-device gesture detection
• – Increased recall from 97% to 99% and reduced false positives by 52% for commercialization across millions of devices
• – Built a low-latency multimodal tutorial agent using Python, asyncio, multiprocessing, and WebSockets to stream audio/video and generate real-time feedback via open-source LLMs

• – Engineered multimodal LLM workflows for mobile proactive agents with concurrent prompt processing
• – Built a real-time proactive mobile audio agent with RAG in Kotlin and Python
• – Optimized a map recognition service using Flask, OpenCV with parallelization to reduce latency by 31%, built React frontend for visualization
• – Built Android communication infrastructure in Java/Kotlin integrating HTTP and WebSockets, reducing duplicate code and accelerating prototyping speed 3×

• – Competed in inter-school and national programming contests
• – Trained 20+ members through weekly Python lectures and group projects

GPA: 3.89