Vellex Computing

Dr. Palak Jain, CEO & Co-Founder, Vellex Computing

Palak Jain is the Co-founder and CEO of Vellex Computing, a startup building ultra-low-power ASIC technology for AI training based on physics-based computing.

Before founding Vellex, Palak held research roles at several prestigious institutions, including Berkeley Lab, GE Research, and Xilinx Research Labs, where she developed machine learning and AI algorithms for edge applications. Her work across these organizations focused on bridging the gap between algorithmic innovation and hardware-efficient deployment, giving her a deep, cross-disciplinary understanding of how AI systems perform under real-world power and compute constraints.

As a researcher at Cal Poly, San Luis Obispo, she conducted foundational research into optimization algorithms using analog computing — work that ultimately laid the technical groundwork for Vellex Computing. This research explored how analog computing architectures could solve complex optimization problems far more efficiently than traditional digital approaches, an insight that now sits at the core of Vellex's technology.

Palak holds a Ph.D. in Electrical and Computer Engineering from the National University of Singapore, where she was a member of the Berkeley Education Alliance for Research in Singapore (BEARS) program, a research partnership between UC Berkeley and Singapore's leading academic and research institutions.

 

Topic:

Ultra-low Power AI Training

 

Abstract:

Vellex Computing (San Francisco, CA) is a Stanford-spinout analog semiconductor company developing ultra-low-power chips for AI training and adaptation. Powered by a decade of foundational research and backed by the U.S. National Science Foundation and Department of Energy, Vellex's technology delivers AI training at a fraction of the energy cost of GPU-based solutions. The company leverages the physics of CMOS analog electronic circuits to drive optimal parameter search in AI models, enabling continual learning, fine-tuning, and on-device adaptation. Vellex holds multiple patents and has active pilots across satellite, drone, and edge AI applications. The company is raising a $1.25M pre-seed round to cross the commercial threshold and accelerate go-to-market, and is seeking partnerships with OEMs and chip makers.

 

Key Technologies Covered:

  • AI Training
  • Analog Computing
  • Edge AI
  • On-Device Training and Adaptation
  • AI Training related Optimization
  • Constrained Optimization
  • Non-Constrained Optimization
  • Self-learning Robots
  • Federated Learning