Neuromorphic computing · Machine learning · Emerging hardware

Artificial intelligence on a biological energy budget.

The Brain Lab builds agile, energy-efficient, and trustworthy hardware for AI. We take findings from neuroscience, turn them into learning algorithms, and map those algorithms onto emerging devices — so intelligence can run on sensors, satellites, and anything else with a battery.

What we do

Our mission is to advance artificial intelligence by exploring the interplay between neuroscience, machine learning, and emerging hardware.

Deep learning's appetite for energy is now the limiting factor almost everywhere it matters: wearables, robots, drones, satellites, implants. Biology solved the same problem on roughly twenty watts. We work backwards from that fact — reading the neuroscience for mechanisms worth borrowing, designing algorithms that survive the translation into analog silicon, and building circuits where memory and computation sit in the same place.

The lab spans fundamental and applied work, from SPICE-level device models to training algorithms and security analysis, with impacts across healthcare, transportation, and defense.

Read the research

Recently

News

8 Sep 2026

Dr. Merkel Presents to the Rochester Museum and Science Center Council

Dr. Merkel joined the Rochester Museum and Science Center Council’s monthly meeting to discuss the state of artificial intelligence, the exciting opportunities ahead, and how the Brain Lab is addressing some of the significant challenges related to AI resource cost and trustworthiness.

21 Aug 2026 Two people standing either side of a screen showing the dissertation defence title slide.

Hagar Hendy Defends Dissertation

Brain Lab and RIT ECE doctoral candidate Hagar Hendy has successfully defended her dissertation, “Time-Domain Circuit Design Techniques for Neuromorphic Computing.” Hagar’s work advances the efficiency and reliability of time-domain computing for edge AI. Congratulations, Dr. Hendy!

5 Aug 2026 Three people standing either side of a projection screen showing the SpikeRFF thesis defence title slide.

Karthik Kumar Defends Thesis on Efficient Spike-Based Learning

Congratulations to Karthik Kumar on successfully defending his MS in AI thesis, “SpikeRFF: Learning Forward with Layer-Local Plasticity and Semi-Hard Negatives.” His work explores self-supervised learning of spiking neural networks without backpropagation. Thank you to Alex Ororbia and Sathwika Bavikadi for serving on Karthik’s committee.

All news

Publications

Read the work.

Papers on memristive learning circuits, spiking network hardware, reservoir computing, and the adversarial robustness of neuromorphic systems.

Browse publications
Join us

We're looking for curious, talented and hard-working people who aren't afraid of the most challenging problems.

Openings for PhD, MS, and undergraduate researchers in computer engineering, electrical engineering, and adjacent fields. Circuit designers, algorithm people, and anyone who wants to work between the two are all welcome. Email a CV and a paragraph on what you'd want to build.

Email the lab