Q & A: Hubertus Franke on Confidential Computing, AI and Academic Collaboration

Q&A

Hubertus Franke on Confidential Computing, AI and Academic Collaboration

Hubertus Franke is an IBM distinguished research scientist and works as an adjunct professor in the Department of Electrical and Computer Engineering at The Grainger College of Engineering at the University of Illinois Urbana-Champaign. His research focuses on the efficiency and security of public and private cloud systems and their applications.

Interviewed by Lilli Bresnahan

What is your research? What do you focus on and why?

Franke: Since joining IBM in 1992, I have always been a “systems person” on a journey. I started out implementing the MPI stack and the gang scheduler for IBM’s supercomputer at the time (IBM SP1/2). It became clear that certain functions needed to be moved out of the operating system (OS), while others needed to be moved in. That raised my interest in operating systems and particularly in Linux, where I made several contributions. That in turn led me to architecture research, where we started integrating accelerators as cores onto the system bus. About 12 years ago, I became more active in cloud system architecture and their control plane. One of my current focus areas is confidential computing, where data is protected not only at rest (storage) and in motion (network), but also during computation. This is accomplished via secure enclaves where memory is encrypted. With the rise of AI and AI accelerators, naturally the question arises: how to expand the secure enclave to include accelerators, essentially forming a distributed yet united cluster of secure enclaves that communicate via IO buses? Finally, with core technology in place, my interests are turning to how to leverage these technologies to deliver higher-level secure services related to machine learning/inference and AI-infused storage.

What is the problem your research addresses?

Franke: Efficiency and security of public and private cloud systems and their applications through optimizations and technologies at the system, operating system and architecture level.

What are  the impacts of your research?

Franke: My research is often the spearhead of creating and investigating advanced technologies for potential integration into IBM products or on how we run our public and private cloud systems. Going back further, many of the innovations I worked on are still in use today, for instance in Linux, such as futexes, scheduling and VM ballooning.

How has AI impacted your research?

Franke: AI is a key driver of my research. AI is in one way a new application, yet its basic blocks are similar to HPC. Many system issues and technologies, such as caching, scheduling and routing, are resurfacing, just at a different granularity and scale.

You show three adjunct professorships. Do you teach or do the same kind of work at all three universities? How does it differ?

Franke: First, I’d like to thank IBM for letting me explore this interest outside. My engagements are different at each university due to their proximity to IBM T.J.Watson. I started teaching OS and computer architecture classes at NYU in 2011 to try it out, and I have been the core instructor for graduate OS classes ever since. Columbia University started out more like IIDAI, where we have joint projects, which triggered interest in seminars on current topics in operating systems research and its intersection with AI. UIUC is solely focused on IIDAI projects that range from operating systems, serverless computing, ML and LLM inferencing and storage architectures.

How has the IBM-IIDAI collaboration benefited you, your work and/or your students?

Franke: I have benefited tremendously from the IIDAI collaboration. I started out under the previous IBM/UIUC collaboration (C3SR) when there was a need to get systems people involved. I learned then that I had access to some of the smartest and most driven students out there. In one way it is an opportunity to learn new things for myself as students explain their vision and insights, but at the same time I can also bring some of my experiences and thoughts into the equation, some of which I would never be able to execute myself due to time constraints. My research is frequently influenced by the projects we are pursuing in IIDAI, such as LLM inferencing, system architectures, content-aware storage and AI systems, as students often come up with innovative ideas that I previously didn’t consider. Finally, the countless joint publications have given my Google Scholar score a significant boost. In return, coming from an industrial/enterprise background, I believe I can provide students with a realistic framework for research exploration and intern/externship opportunities.

Anything else you would like to add?

Franke: 33+ years at IBM have literally flown by. Time is precious; make the most of it. I still stay in touch with many U. of I. students that have graduated.


Share this story

This story was published June 23, 2025.