About
As chip designers, we lived with a cost that never shows up in a schedule or a budget. Every project puts enormous mental, emotional, and physical pressure on the team, from the newest engineer to the leadership, and much of that effort goes into repetitive work that has little to do with creativity or design. Even an inventive team can lose to one that simply out-grinds it.
We never accepted that this was just how chips get made. What drew us to this field was the joy of invention: finding a new circuit or method that improves performance or solves a hard problem. Long before large language models, we worked to bring that back. Now it is within reach. We are building AI that takes on the laborious work and removes steps that exist only because people had to run the flow by hand. Freed from the grind, engineers can return to inventing, think bigger, and build chips they would not dare to imagine today.
If you want to invent again, join us.
Founding Team
Babak has spent 20 years building silicon and AI technology. He founded startups in generative AI, autonomous vehicles, silicon, and human motion modeling, built and led the engineering group at Quantenna on its path to IPO, and served as a senior consultant to startups later acquired by Cadence and a major technology company. Earlier, he held senior roles at Qualcomm and Broadcom and research positions at IBM Research and NEC Labs America. Babak holds a doctorate from Tampere University, Finland, and a Ph.D. in Electrical Engineering from Columbia University.
Mingoo is Associate Professor of Electrical Engineering at Columbia University, where he chairs the Computer Engineering Program. His research focuses on SRAM in-memory computing, and his group developed XNOR-SRAM and has prototyped more than 40 test chips. He has published more than 150 papers and advised more than 20 Ph.D. graduates, now at companies including NVIDIA, Apple, Google, TSMC, and Intel. He received the NSF CAREER Award and the Qualcomm Faculty Award. Mingoo holds a Ph.D. in Electrical Engineering from the University of Michigan.
Matthew’s recent work spans machine learning infrastructure and edge AI deployment. He brings 20 years of analog and mixed-signal IC design: high-speed data converters and clocking systems in advanced CMOS nodes at Broadcom, LSI, and Agere Systems, shipped in high-volume storage and networking products, and touchpad silicon at Synaptics. Matt holds a B.S. from Caltech. and a Ph.D. in Electrical Engineering from Johns Hopkins University.
