I build robots that learn to move.

Projects

BDX-R

ISAACLAB // RL // CAD // 3D PRINTING // PYTHON

A custom bipedal robot inspired by Disney's BD-X Robots. I reverse engineered the BD-X to build my own, then trained and deployed a walking RL policy from IsaacLab to real hardware.

View Project

Stompy

REINFORCEMENT LEARNING // ROBSTRIDE // 3D PRINTING // CAD

A fully 3D-printed bipedal robot built and taught to walk in 7 days for a Bambu Lab challenge, trained with RL and deployed to the real hardware.

View Project

Booster T1 RL Locomotion

ISAACLAB // MUJOCO // PYTHON // PYTORCH

Migrating Booster Robotics' deprecated Isaac Gym pipeline to IsaacLab, and building custom RL walking policies for the Booster T1 humanoid.

View Project

Booster T1, Kicking

ISAACLAB // REINFORCEMENT LEARNING // SIM2REAL

Training a Booster T1 humanoid to kick a ball, from IsaacLab simulation to a real RoboCup field.

View Project

Dancing with Booster T1

BEYONDMIMIC // ISAACLAB // MOTION RETARGETING

Implementing BeyondMimic on the Booster T1 to retarget motion capture dance sequences into a real RL policy.

View Project

Robstride Actuators Control

PYTHON // CAN BUS // MIT CONTROL

Open-source Python driver for controlling Robstride motors over CAN bus using MIT Control Mode, now used by thousands of builders.

View Project

🏆 HACK2026, 1st Place

LEROBOT // NVIDIA GR00T // IMITATION LEARNING // GPS NAV

Won ETH Robotics Club's HACK2026 against 500+ applicants. Built a humanoid robot that autonomously navigates outdoors and handles objects, targeting Europe's labour shortage.

View Project

Coming Soon: Bipedal Robotic Character Design Internship

CHARACTER DESIGN // BIPEDAL ROBOTICS

Currently interning on bipedal robotic character design. Full writeup once I'm able to share details.

Coming Soon

About

I'm a robotics engineer focused on reinforcement learning and sim-to-real transfer, taking walking policies from simulators like IsaacLab and MuJoCo and getting them running reliably on physical actuators. My Bachelor End Project at TU/e, a from-scratch bipedal robot (BDX-R), was graded 9/10 and walked in the real world by the end of it. I also document my builds on YouTube, where my project videos have picked up over 100K+ views.