LATEST IN THE NEWS

December 14, 2023

Vellex Computing Completes NSF I-Corps Customer Discovery Program

Vellex Computing completed the National Science Foundation's I-Corps program through the I-Corps Mid-South Hub, conducting 125 customer discovery interviews across energy, manufacturing, and electric vehicle sectors to validate the market demand for on-device AI and real-time edge intelligence.
Read Vellex News
Vellex Logo Small
September 15, 2023

Vellex Computing Pitches at 2023 PG&E Innovation Pitch Fest

Vellex Computing presented its analog computing technology at the 2023 PG&E Innovation Pitch Fest in San Ramon, California, demonstrating grid optimization applications to PG&E leaders and energy industry partners.
Read Vellex News
Vellex Logo Small
June 9, 2023

Vellex Computing Co-Founder Named 2023 Activate Berkeley Fellow

Dr. Palak Jain, CEO and co-founder of Vellex Computing, was named a Cohort 2023 Activate Fellow at the Activate Berkeley Community at Lawrence Berkeley National Laboratory's Cyclotron Road, one of 46 fellows selected from a record pool of 832 applications.
Read Vellex News
Vellex Logo Small

OUR BLOGS

May 28, 2026

Analog vs. Digital: The Ultimate Guide to Choosing the Right Tech

Vedant Wakchaware
Analog vs. Digital technology is a decades-old debate, but which one is actually better for the future of tech? In this comprehensive guide, we break down the core differences between continuous analog waves and discrete digital steps. Discover the unique pros and cons of each system, learn which architecture is best suited for your specific industry—from professional audio to cutting-edge AI—and explore how modern hardware engineers are blending both into powerful, ultra-efficient Hybrid (A/D/A) systems to overcome the massive energy limitations of today's devices.
May 8, 2026

The Hidden Economics of AI: Why Tokens Are Costing Millions in Training and Usage

Vedant Wakchaware
Tokens are the invisible currency of artificial intelligence, and processing them carries a staggering hidden cost. Why does training a frontier model require gigawatt-hours of electricity, while developers face unexpected API bills? Step inside this foundational metric to decode these massive economics. This deep dive breaks down the physical infrastructure required to process trillions of data fragments—from specialized GPU clusters to liquid cooling taxes. Discover the invisible multipliers inflating your usage, and learn comprehensive optimization strategies like smart routing and prompt caching to drastically reduce your AI costs.
April 30, 2026

A Detailed Guide to Federated Learning on Edge Devices

Vedant Wakchaware
While on-device training secures user privacy, it unintentionally traps intelligence, forcing every edge device to learn the exact same lessons from scratch. How do we build a collaborative "hive mind" without exposing raw data to the cloud? The answer is Federated Learning. This comprehensive guide explores the decentralized paradigm of bringing the model to the data, detailing how devices evolve together by sharing abstract mathematical updates. Dive into the 5-step federated architecture loop and discover how cryptographic shields like Secure Aggregation and Differential Privacy prevent data extraction. Learn how advanced algorithms overcome severe bandwidth constraints and hardware disparities to power the next generation of secure, collective AI.

OUR PUBLICATIONS

April, 2026

Automated Synthesis of Hardware-implementable Analog Circuits for Constrained Optimization

Sachin Khoja; Kamlesh Sawant; Palak Jain; Sairaj Dhople; Jason Poon
December, 2024

A hybrid-computing solution to nonlinear optimization problems

Kamlesh Sawant; Dillon Nguyen; Alex Liu; Jason Poon; Sairaj Dhople
Published in IEEE Transactions on Circuits and Systems I - Regular Papers, vol. 71, no. 12, pp. 6555-6568, Dec. 2024
May, 2022

Real-time selective harmonic minimization using a hybrid analog/digital computing method

Jason Poon; Mohit Sinha; Sairaj V. Dhople; Juan Rivas-Davila
Published in IEEE Transactions on Power Electronics, vol. 37, no. 5, pp. 5078-5088, May 2022