I work with a great team at a young robotics start-up, Dexterity. We are a full-stack team -- spanning work in mechanical design, mechatronics, controls, computer vision, user interfaces, cloud/on-premise hybrid software, etc. It's a fun environment, solving real problems faced by our clients.

The majority of my own work is in computer vision, machine learning and software infrastructure -- helping our robots perceive the world and making it easier for our engineers to make the robots do interesting things.

Building dialog systems for the Google Assistant.

Assisting instruction of introductory signal processing course for electrical engineering undergraduates. Developed material for and held weekly review/problem sessions. Aided student understanding in small group settings during office hours twice a week.

Academic support for EE 102B, one of the core signal processing classes for an electrical engineering undergraduate at Stanford. I held office hours, review sessions, and created supplementary material to aid in the students' understanding of the material. Course topics include sampling and reconstruction, FIR filters, discrete Fourier transforms, Laplace transforms, feedback control systems, Z transforms and their applications, and IIR discrete-time filters.

I developed large-scale simulations of biologically-plausible spiking neural networks for robot control (specifically operational space control) using the neural engineering framework and Nengo, and connected the simulator to a physical robotic arm, controlling it in real-time. My work was part of a demo at the Neural Information Processing Systems (NIPS) conference in December 2013.