About the Lab

Our ultimate goal is to make AI both accessible and sustainable, and far less dependent on large-scale AI data center infrastructure.

What We Do?

The Resource Efficient AI Lab (REAL), directed by Reza Rawassizadeh, works on moving AI off large cloud infrastructure and onto the edge. We develop optimization methods such as model pruning and task-specific pruning, quantization, compression, and energy-aware training, so that capable models can run on resource-constrained hardware, down to phones, mobile robots, and wearables.

A central focus is digital health. By running models directly on wearables, robots, and mobile devices, we enable health monitoring and behavioral insights that work without a constant network connection and keep sensitive data in the user’s hands. This makes AI-driven health tools cheaper to deploy, easier to scale, safe from third-party access, and available to people beyond well-resourced clinical settings.