Software engineer on the options team

• Worked under Stuart Russel to evaluate reinforcement assistance learning through sandbox games. Trained agents using RLlib and Ray to observe a human’s actions and assist appropriately. Built infrastructure to communicate between PyTorch and Minecraft.

• Designed, implemented, and pushed to production a new way to manage remote kernels in JupyterLab, extending complex existing Python and Javascript libraries to support new features.

• Enhanced security, performance, and customizability. Increased startup speed by 20x.

• Proposed and implemented a Jupyter Enhancement Proposal and multiple PR’s in open source adding widely demanded features.