About

I'm an engineering manager and wireless systems engineer in Seattle. I joined Project Kuiper, now Amazon Leo, in March 2020 as one of its first hundred employees. Before that I did machine-learning research for wireless at UT Austin. I finished my M.S. there in 2019 and left the PhD program early to join Kuiper.

At Leo I've worked on the modem at every stage, from simulation to orbit. I built the MATLAB simulation of the waveform and architected the modem verification infrastructure. I led PHY bring-up across seven chip iterations and refactored the customer-terminal downlink firmware into production code. Then I led analysis of the PHY data the satellites send back from orbit. I watched the first prototype satellites launch on an Atlas V from Cape Canaveral.

Today I lead a modem firmware team of seven engineers. One of our four charters is AI-first, agentic test infrastructure, and I'm driving the team's use of AI tools for test generation, log analysis and firmware development. That work is why I'm looking at what comes next: helping bring frontier AI to people the way satellite internet reached them, from early prototype to product at scale.

My research was ML-assisted LDPC decoding at UT Austin: neural networks that learn LLRs from quantized samples, and belief propagation unrolled into a trainable network. At Intel Labs I built a deep-learning traffic classifier. Earlier internships took me to Astranis, Apple, Microsoft and MIT Lincoln Laboratory, and I built CubeSat electronics in the Michigan eXploration Laboratory. The drawing register has the research and the projects I build outside work.