I build interactive systems for autonomous machines, simulation, and engineering education—turning difficult workflows into experiences people can see, test, understand, and refine through direct interaction.
DroneDream is a local-first PX4 and Gazebo tuning platform that turns control-parameter search into a visual, repeatable engineering workflow. It connects experiment setup, simulation, comparison, and evidence so users can understand each tuning decision, revisit results, and move from trial and error toward reliable autonomous flight.
RaceCar Lab is an interactive formula-car engineering environment with a code-generated 3D vehicle, component studies, formulas, diagnostics, and bilingual learning paths. Its McLaren-inspired presentation reflects my support for McLaren and Lando Norris, while my girlfriend supports Ferrari and Charles Leclerc—turning our rivalry into a way to explore racing engineering.
CoRe-VQA is a collaborative multi-agent framework for implicit video question answering. It separates clue scouting, relation judging, consensus voting, and confidence-gated memory refinement into specialized roles. Prepared for AAAI submission, it improves hidden-evidence I-VQA accuracy from 55.14% to 63.59% across traffic-video and transfer benchmarks.
NAVA is an AI biological-rhythm app that connects mobile health records, HealthKit data, Supabase services, and rhythm modeling into a daily wellness companion. I lead the app development and work as an algorithm-team contributor, helping refine the rhythm engine, validate multi-day signals, and turn model outputs into clear product interfaces so daily feedback stays lightweight, personal, useful, and actionable.
I am a senior undergraduate student in Electrical and Computer Engineering at Zhejiang University. During my junior year, I spent an academic year as an exchange student at the University of Illinois Urbana-Champaign (UIUC). Beyond the classroom, I am currently conducting research in Bruce Yu's group on human action recognition from UAV-view video, with a particular interest in understanding human behavior and temporal patterns from aerial observations.
Previously, I completed a year-long Student Research Training Program in Hanzhi Ma's group (National SRTP). My research focused on high-speed signal-channel modeling using machine learning and multi-objective optimization, and the project received an Excellent rating at its final evaluation. During the summer after my sophomore year, I also worked in Jixing Cao's group on YOLOv11-based detection of tunnel fires and vehicle collisions.
I am also leading the app development of NAVA, an AI biological-rhythm app, while supporting the algorithm team in refining rhythm-model iterations and connecting them to real product interfaces.
I completed RaceCar Lab, an interactive website that gives Formula racing beginners a structured path through vehicle architecture, core engineering principles, component analysis, and hands-on 3D exploration, making a complex subject easier to understand.
During my month at Fudan University’s Zhangjiang Institute, guided by Suli Ye, I developed five AI × real-estate systems for land discovery, site planning, home design, rental services, and silver-community living across the property lifecycle and resident care.
I formally completed the ECE 498 DroneDream project, consolidating our PX4 and Gazebo work in an interactive platform that makes autonomous-flight tuning visual and repeatable and connects experiment setup, simulation, comparison, and evidence.
I was named to the Fall 2025 Dean’s List at The Grainger College of Engineering, University of Illinois Urbana-Champaign.
I published the Autonomous Tracking Drone project developed for ECE 484, together with a video demonstration of the system.
Westlake University interviewed me about why I chose its EIE Summer School and what I learned there. I reflected on its international, cross-disciplinary environment, which encouraged me to leave my comfort zone, exchange ideas, and learn across fields.
Our National SRTP project received an Excellent evaluation in a cohort of 24 participating teams.
Our summer research project received both the Best Summer Research Project award and the Best Summer Research Poster award.
