SAFE-AI: The 1st Workshop on
Systems & Architectures For
Encrypted AI
 
31st March 2025
Collocated with ASPLOS 2025, Rotterdam, Netherlands

About

SAFE-AI: The first workshop on Systems and Architectures for Encrypted AI

This workshop on Encrypted AI will explore cutting-edge techniques for accelerating and democratizing privacy-preserving AI, focusing on Fully Homomorphic Encryption (FHE) and Secure Multi-party Computation (MPC). Designed in two parts, the workshop will first host sessions with invited speakers from academia and industry, who will share the latest advances in FHE for AI, real-world applications, and challenges in scaling encrypted AI. The workshop will cover a broad range of topics on systems and architectures for encrypted AI including acceleration techniques across the computing stack, compilers, tools, DSLs, and applications for encrypted AI.

The second part of the workshop is a hands-on tutorial centered on the Cinnamon Framework for scale-out encrypted AI, providing participants with a comprehensive understanding of essential RTL design principles, architecture, compilers, DSL, and encrypted AI applications and models. Attendees will learn practical skills for creating, exploring, and deploying encrypted AI workflows, leveraging Cinnamon’s modular approach to tackle the complexity of FHE on multi-chip systems. By the end of the workshop, participants will be equipped with tools and knowledge to scale AI models securely, integrate FHE into diverse AI applications, and bridge the gap between theoretical encryption techniques and practical, high-performance AI solutions. This event is ideal for researchers, engineers, and industry professionals invested in privacy-preserving AI.

Program

Tentative Schedule

Time Event
8:30am Coffee
9:00am Talks
10:30am Break
11:00am Talks
12:30pm Lunch
2:00pm Tutorial
3:30pm Break
4:00pm Tutorial
5:30pm End

Talks

Talks featuring speakers across acadmeia and industry on the latest exciting developments in encrypted computing.

Time Speaker Title
TBD TBD TBD

Tutorial

A Tutorial on the State-of-the-art Cinnamon framework for scale out encrypted AI.

Organizers

Siddharth Jayashankar, Wenting Zheng, Dimitrios Skarlatos

Carnegie Mellon University

Contact

Dimitrios Skarlatos (dskarlat@cs.cmu.edu)