Robotics Engineer · M.S. Robotics, University of Michigan
I am a Robotics graduate student specializing in robotic controls, manipulation, and simulation.
Trained a velocity-conditioned walking policy for a 12-DoF Barkour quadruped using PPO in Brax / MuJoCo MJX, running thousands of parallel environments on GPU. Designed the reward function for body-frame linear and yaw velocity tracking, and used an MLP policy over an observation history to output joint targets at 50 Hz.
Implemented the core of a Neural Radiance Field in PyTorch to reconstruct the Blender synthetic “Lego” scene from multi-view images. Built pixel-to-ray generation from camera intrinsics and extrinsics, stratified sampling along rays, and a differentiable volume renderer producing RGB, depth, and opacity maps. Extended the pipeline with hierarchical coarse-to-fine sampling to recover finer geometry.
Built a high-fidelity simulation framework for networked multi-robot systems using PX4 and Gazebo, coordinating multiple x500 quadrotors over a shared network for cooperative autonomous survey missions.
Designing an automated robotic workcell for end-to-end server disassembly at the CMU Robotics Institute (Matt Lab), part of an industry-sponsored e-waste recycling initiative integrating perception, manipulation, and mechanical subsystems.
Developing an autonomous surface vehicle (ASV) for the PEP 26 Autonomy division, integrating GPS/GNSS waypoint navigation and closed-loop heading and speed control via ArduPilot / MAVLink. Fusing data from three redundant IMUs for robust attitude estimation and fault tolerance during open-water field operations.
Developed a high-fidelity ROS 2 / Gazebo simulation of a mobile manipulation platform (Clearpath Husky UGV + UR3 arm + Robotiq gripper). Authored custom joint-space trajectory controllers and fused simulated LiDAR point clouds with wheel-encoder odometry for real-time localization; architecture designed to support future autonomous grasping pipelines.
Led the embedded controls architecture for GHOST, UIUC’s first electric motorcycle team. Developed PID-based closed-loop torque control firmware for a 45 kW PMAC motor on a 103.6 V drivetrain; aggregated encoder, IMU, temperature, and BMS telemetry over CAN bus into a real-time operator dashboard.
Derived and implemented an EKF for fusing GPS position fixes with INS inertial measurements, enabling robust 6-DOF state estimation under simulated sensor noise. Formulated nonlinear process and measurement models with analytical Jacobians, modeled gyroscope and accelerometer bias drift, and benchmarked localization accuracy against ground-truth trajectories.
Built a LiDAR SLAM pipeline on an EarthSense TerraSentia agricultural UGV, processing raw point cloud data with a RANSAC-based line-fitting algorithm for feature extraction. Fused scan-derived landmarks with dead-reckoned odometry to estimate robot pose and publish a globally consistent 2D occupancy grid to the ROS /map topic.
View on GitHub →
Michigan Robotics — DASC Lab
Carnegie Mellon University Robotics Institute — Matt Lab
Whisper Aero
USACE — Robotics for Engineer Operations
B.S. in Mechanical Engineering from the University of Illinois Urbana–Champaign and current M.S. Robotics student at the University of Michigan. My experience building and working with real robotic systems has prepared me to take on diverse robotics projects.
Updated resume, August 2026.
Download Resume