At Alignerr, I design MuJoCo robotics simulation tasks to train Anthropic's next-generation AI models. Four tasks involving wheeled mobile robot control, manipulation, and computer vision were accepted and delivered directly to Anthropic.
For my Reinforcement Learning – Mobile Robot project, I designed a custom robot and developed a realistic MuJoCo simulation for direct Sim-to-Real transfer. With my peers, I assembled the robot and used a learned policy to climb dirt walls; I’ve also designed and 3D-printed prototypes in SOLIDWORKS and worked with ABB and FANUC manipulators.
