CV - Ying-Chun Lee

Education

M.S. in Electrical and Computer Engineering

University of Washington, Seattle, WA

Sep 2025 — Mar 2027 (Expected)

GPA: 3.95 / 4.0

B.S. in Electrical Engineering

Yuan Ze University, Taoyuan, Taiwan

Sep 2020 — Jun 2024

GPA: 3.73 / 4.0 · Major: 3.94 / 4.0

Experience

Amazon — Seattle, WA

PRIOR Team, Allen Institute for AI (Ai2) — Seattle, WA

Enabled real-time imitation-learning deployment across 3 robotic platforms (Franka FR3, SO-100/SO-101 via LeRobot, YAM), cutting data-collection time from 4 hours to ~100 minutes per 100 episodes
Fine-tuned and deployed π₀, π_0.5, π₀-FAST, ACT, SmolVLA, and Diffusion Policy on real hardware, adding joint-position/velocity action support and real-time inference on the Franka FR3
Built an end-to-end data collection pipeline covering calibration, recording, validation, and LeRobot-compatible dataset conversion for scalable policy training
Engineered GELLO teleoperation for the Franka FR3 and YAM; integrated Meta Quest 2 Cartesian-delta control for the FR3 via Polymetis and for the YAM via direct SDK interface

University of Washington — Seattle, WA

Architected a distributed pipeline spanning 400K+ repositories, reducing collection time by 87% (5 hours to 40 mins) with hybrid GitHub GraphQL/REST strategy, Redis, Celery, and multi-token rotation
Engineered a secondary validation layer to sanitize raw API data, ensuring integrity by detecting legal blocks (HTTP 451) and correcting metric inconsistencies for ~2% of the dataset
Delivered a tiered scoring system and React interface with instant filtering (<100ms), client-side pagination, and URL state persistence

Husky Robotics Team, University of Washington — Seattle, WA

Delivered a real-time perception stack for an autonomous Mars rover, enabling pose-aware navigation and ~90% object-detection accuracy by integrating OpenCV, ArUco solvePnP, and OWL-ViT
Built a fine-tuning pipeline for OWL-ViT to accurately detect three designated objects in the University Rover Challenge (URC), improving recall on domain-specific targets
Integrated Intel RealSense depth sensing into the perception stack, enabling the rover to estimate object distance in real time for autonomous approach and manipulation
Architected camera calibration and perception-to-navigation integration in C++ for low-latency real-time control

Chang Chun Group — Information Center, Taipei, Taiwan

Created an end-to-end imitation-learning training platform for the ALOHA VX300S robotic arm, achieving 100% grasp success in simulation across a 35–65 cm workspace
Developed a ROS2–Isaac Sim digital twin and Transformer-based ACT++ control pipeline with a three-stage trajectory planner
Implemented automated conversion to standard ACT++ HDF5 datasets for scalable training

DRES Lab, University of Illinois at Urbana-Champaign (UIUC) — Prof. Yih-Kuen Jan & Prof. Chih-Yang Lin

Researched medical image recognition techniques; implemented U-Net, Mask R-CNN, and DeepLab V3+ for automated shoulder soft tissue segmentation from ultrasound images
Published first-author paper in IEEE Access (2024)
Developed a deep learning model combining VGG and ANN for spinal cord injury (SCI) classification from shoulder ultrasound images
Selected as Scholarship Recipient through Taiwan Ministry of Education Overseas Internship Program (2022 & 2023)

Projects

ROS & Gazebo — Yuan Ze University

Built and tested an autonomous four-wheeled delivery robot using ROS Noetic, Gazebo, C++, and LiDAR
Implemented gmapping for SLAM-based mapping, amcl for localization, and costmap for real-time path planning and obstacle avoidance

Raspberry Pi 4 & Flask — Yuan Ze University

Developed a web-controlled door lock system using Raspberry Pi 4, RFID reader, and a Flask web interface
Resolved electromagnetic lock overheating by implementing an auto-lock feature to prevent prolonged current flow

Computer Vision — Yuan Ze University

Developed a real-time gesture recognition and background replacement system achieving 99% accuracy across five gestures
Extracted 522 body landmarks via MediaPipe; optimized four classifiers including Random Forest and Gradient Boosting

Teaching

Yuan Ze University

Assisted professor with coursework management, grading, and exam logistics
Provided one-on-one support for students on LC3 assembly language, two's complement, and binary operations

Yuan Ze University

Taught Arduino development and C++ fundamentals to university students and conducted workshops for high school students

Publications

Abhay Deshpande, Maya Guru, Rose Hendrix, Snehal Jauhri, Ainaz Eftekhar, ..., Ying-Chun Lee, ..., Georgia Chalvatzaki, Dhruv Shah, and Ranjay Krishna, "MolmoB0T: Large-Scale Simulation Enables Zero-Shot Manipulation," arXiv:2603.16861, 2026. arXiv
Yejin Kim, Wilbert Pumacay, Omar Rayyan, Max Argus, Winson Han, ..., Ying-Chun Lee, ..., Ali Farhadi, Dieter Fox, and Ranjay Krishna, "MolmoSpaces: Large-Scale Open Ecosystem for Robot Manipulation and Navigation," Submitted to RSS 2026. arXiv
Shirui Chen, Cole Harrison, Ying-Chun Lee, Angela Jin Yang, Zhongzheng Ren, Lillian J. Ratliff, Jiafei Duan, Dieter Fox, and Ranjay Krishna, "TOPReward: Token Probabilities as Hidden Zero-Shot Rewards for Robotics," Submitted to ICML 2026. arXiv
Ying-Chun Lee, Chih-Yang Lin, Chia-Chun Hsiao, Pu-Chun Mo, Jiaqi Guo, and Yih-Kuen Jan, "Using Deep Learning-Based Methods for Automated Segmentation of Soft Tissues from Shoulder Ultrasound Images," IEEE Access, Vol. 12, pp. 111,481-111,492, July 2024. DOI

Awards

Selected to conduct research at UIUC

Yuan Ze University

Yuan Ze University — 1st out of 10 in Finals

Taiwan — 8th out of 36 teams

Skills

Programming

Python C C++ JavaScript HTML CSS SQL MATLAB

Robotics & AI

ROS2 PyTorch TensorFlow Keras OpenCV NumPy Matplotlib PIL NVIDIA Isaac Sim

Web & Cloud

React Node.js Flask Docker Kubernetes AWS GCP Redis Celery

Data & Analysis

pandas SciPy Jupyter Git