George Cui / Suzeyu Cui | Mechanical Engineering, Robotics, and Research

Rutgers SCAR VEXU

Competition robotics design and driving work during the 2025-2026 season, including more than 300 CAD hours and full robot design responsibility.

HackPrinceton 2026

Built with close friends at HackPrinceton 2026: a full-stack voice-based interview coaching platform with resume-driven personalization, company research, role-specific behavioral questions, AI evaluation, and privacy-aware webcam delivery analysis.

Stack: React, TypeScript, FastAPI, PostgreSQL, OpenRouter, ElevenLabs, MediaPipe, and Clerk.

FIRE Robotics

Captain and Head of Finance for a 30+ member robotics team with technical, fundraising, and outreach responsibility. The team secured $3,200 in sponsorships and organized outreach for more than 1,000 attendees.

LHS Rocketry Team 1

Co-captain of a school aerospace organization that grew into a 50+ member technical team with launches, mentorship, and community demonstrations.

Teaching 4 Charity

Regional Executive Director for a student-led tutoring and service organization supporting local outreach and educational initiatives in Uganda. The organization raised roughly $10,000 for educational programs.

Livingston Huaxia Volunteer Club

President of the Livingston Huaxia Chinese School Volunteer Club, focused on volunteer operations, faculty communication, service-hour accountability, and community event support.

Aresty Summer Science Program

Selected as the sole researcher from 56 applicants for Rutgers Aresty's most popular engineering track. Spring-Summer 2026 research focuses on decoding flapping-insect flight mechanics with computer vision, image segmentation, and reinforcement learning, then translating those principles into autonomous drones.

• Studying complex biological flight patterns at the intersection of biology and engineering.
• Developing insect-inspired drone concepts with CAD, computational fluid dynamics, and multiphysics simulations.
• Building toward physical prototypes through 3D printing and mechatronics.

Eye-Gaze-Controlled Robotics

Assistive robotics research exploring gaze direction as an input signal for selecting or guiding robotic movement when conventional input devices are difficult to use.

Optimizing Reinforcement Learning Using Failure Data

First-author IEEE MIT URTC paper introducing DQN-DMM, a failure-aware Deep Q-Network that uses a modified Gaussian penalty to guide agents away from historical failure points. The approach improved learning speed and stability across CartPole, shower-temperature control, and a real-world robotic-arm task.

• CartPole: optimal performance in roughly 14,000 steps instead of 20,000 for baseline DQN.
• Shower control: trained in roughly 5,000 steps while baseline DQN still struggled after roughly 14,000.
• Robotic arm: score near 60 in 175 episodes versus a baseline maximum of 8 after more than 4,000 episodes.

Read my IEEE paper

PFAS Water Treatment Research

Environmental engineering research with Dr. Yang Deng on how granular activated carbon and iron ions can help clear PFAS from water.