I’m currently the CEO & Co-Founder of Gradient Robotics. I’m a robotics engineer and led engineering and product teams at K-Scale Labs (YC W24). I’ve worked on projects in mechanical, electrical, design, software, ML, marketing, and manufacturing.
I love building cool projects with cool people. Shoot me a message if you’re working on something cool!
K-Bot is an open-source, developer focused humanoid platform running RL-based locomotion and VLA manipulation.
I’m the main creator of the robot and led the engineering and product at K-Scale across the stack — ML training library, operating system, hardware design, and manufacturing from prototyping to production in less than 6 months.
K-Bot is the United States’ first consumer & open-source humanoid robot. In the July 2025 launch, we sold out the first 150 production units for USD$2M+ to developers and top labs from Google Deepmind, OpenAI, and Stanford.
See: Product page (Website), Software stack (Youtube), Hardware stack (YouTube)
Independent coverage: The Information (May 2025), Late Post (June 2025), HN front page (May 2025)
Following the success of the Zeroth-01 robot, I’m working on bringing the robot to mass production with more developer tools and capabilities. Currently, the prototypes have been used in Stanford’s CS234 Reinforcement Learning course.
The robot is sub-$1k with open-source SDK & sim for RL/VLA. Currently with 20,000+ on the public waitlist.
https://www.youtube.com/watch?v=O6zqIltJcVw
Zeroth-01 is the world’s smallest 3D printable and open-source end-to-end humanoid robot for $350 BoM. I and friends (Kelsey and Denys) designed and built the stack from RL training and the operating system, designed the hardware, and launched it. The open-source community now has 5000+ members with over 700+ stars on GitHub, and used by educators, developers, and labs all over the world!
https://www.youtube.com/watch?v=pPCiDQ4IMRU
Industrial robot visual programming, testing, and deployment platform that allowed non-programmers to construct industrial robot behavior trees through a graphical interface, allowing testing in simulation, debugging, and deployment onto real robots.
We shipped features to partners and clients including Mecademic Robotics.
I built controllers, planners, inverse/forward kinematic/dynamics algorithms, and action client/server framework from scratch with ROS1 with C++ for the Kinvoa Gen3 robotic arms for dynamic grasping.
Built controllers: kinematics and dynamic control, redundancy resolution, force control and contacts, grasping. Built planners: trajectory generation, potential fields, collision detection and avoidance.
Student @ McGill University (2023)
I built an autonomous vehicle with the NTU team for the F1TENTH autonomous racing challenge. I designed the software architecture for ROS topics, vehicle state, dynamics, and navigation. I implemented global trajectory planning and local avoidance algorithms in ROS2, integrating hardware and software for planning and controller.
Student @ NTU Singapore (2024)