Mandy Terrill – University of California, San Francisco

In 2019, a cyberattack rattled the University of California, San Francisco, exposing just how vulnerable even the most sophisticated research institutions can be. For most people outside the walls of academic IT, that kind of event might have faded into memory. For Mandy Terrill, it became the starting point of a mission. Hired a few years later as assistant CIO of Research, she was brought in to make sure nothing like that could compromise the university’s science again — and to build something UCSF had never had before: research infrastructure strong enough to match the ambition of the work happening inside its labs.

Mandy Terrill | University of California, San Francisco | Assistant CIO of Research

UCSF is not a typical university. It has no undergraduates. Every student walking its halls is already pursuing an advanced degree, whether in the school of medicine, nursing, dentistry, or pharmacy, and the institution operates simultaneously as a leading academic medical center and a hospital system. That dual identity means Terrill’s team isn’t just protecting data for the sake of a classroom exercise — it’s safeguarding patient outcomes, federally funded grants, and discoveries that could shape the next generation of medicine. It’s a responsibility that touches nearly every corner of the institution, from a lab chasing an early breakthrough in cancer research to a clinical unit relying on that same secured network to keep patient information safe.

Building a home for high-performance computing

The centerpiece of Terrill’s work is the Facility for Advanced Computing, a new research infrastructure built to give UCSF’s scientists the horsepower they need without sending their data anywhere else. Before it existed, analyzing a large dataset might mean stringing together hundreds of laptops and waiting years for results. Now, with high-performance computing and machine learning tools in place, researchers can run in days what once took a decade.

That speed matters because the science behind it is rarely simple. Grants from agencies like the National Institutes of Health and the National Science Foundation come loaded with their own compliance requirements, and much of the research touches human subjects or animal studies governed by strict review boards. Terrill’s team designed the system with a dual environment and a partition between clusters, so joint faculty can run large-scale jobs across every node when they need extra capacity, without ever compromising the security walls built around sensitive data.

“This is a unique configuration, and it helps us be one collaborative team,” Terrill says.

More than 100 individual compliance controls now govern how that data moves and where it lives, a framework Terrill’s team built to satisfy federal regulators and reassure faculty that their research, much of it years in the making, would remain protected. UCSF recently completed an audit of the system, using the results to identify where those controls could be strengthened even further.

Earning trust across UCSF

None of it would matter without buy-in, and Terrill has spent much of her tenure building it one relationship at a time. UCSF convened a task force in 2020 to rethink how IT could better serve research, and when Terrill arrived in 2022, part of her charge was to place dedicated leads inside each mission area of the university so that IT decisions could be shaped by the people actually doing the science, not made in isolation from it.

“We wanted to understand the uniqueness of each mission area and then bring that back into IT,” Terrill says.

That approach meant sitting down with faculty across departments, listening to what they needed to compute at scale, and then making the case up the chain. Terrill and her team pitched the investment directly to the chancellor, framing it not as a cost but as a competitive advantage for a university trying to recruit and retain top scientific talent in an increasingly AI-driven research landscape.

Weighing the cloud against building on campus

Before committing to that investment, Terrill’s team ran a request for proposals to compare building the infrastructure on premises against paying for on-demand cloud computing hosted elsewhere. The market research made the answer clear: owning and operating the system themselves would be far less expensive over time than outsourcing it, while also keeping sensitive research data inside UCSF’s own walls rather than sending it to outside vendors or national labs.

That decision has already paid dividends. Building on premises has saved the university a significant amount of money, freeing up resources that faculty can put back into the research itself, and giving scientists a cost-effective way to pursue ambitious projects without looking outside the institution. It also sidesteps a logistical headache: transferring large volumes of protected data offsite, a process that is often slow, costly, and risky from a compliance standpoint.

Much of that work has been done in partnership rather than alone. UCSF has worked closely with Cambridge Computing to help build out the cluster, while its long-standing relationship with the Gladstone Institutes has proven especially valuable as the university expands its computing footprint.

“Our partner at Gladstone has been huge — they’ve been a huge asset,” Terrill says.

A team built for growth

Terrill’s own team has grown alongside the infrastructure it supports. What started as a one-person operation now numbers three, with two additional engineers recently hired to help keep pace with demand. That growth reflects a broader reality: as more UCSF faculty push into large language models and other frontiers of artificial intelligence research, the appetite for computing power shows no signs of slowing down.

Much of that future growth depends on continued investment, and UCSF, like many research institutions, relies heavily on private funders and philanthropic support to keep expanding its capabilities. Terrill sees fundraising not as a separate task from her technical work, but as an extension of it — a way of making the case that secure, well-built infrastructure is itself a form of scientific progress. Recruiting and retaining faculty is part of that pitch, too; in a landscape where top researchers can take their work almost anywhere, having a fast, secure, and well-resourced computing environment on campus has become a competitive edge in itself.

“This helps us advance science in an accelerated way,” Terrill says.

For an institution built entirely around graduate and postgraduate discovery, that acceleration carries real weight. Every hour saved on a computing job is an hour a researcher can spend chasing the next hypothesis, the next grant, or the next breakthrough that might change how a disease is treated. Terrill’s work rarely makes headlines outside the university, but inside it, the impact is felt every time a scientist logs onto a secure cluster and gets an answer in days instead of years. That, in the end, is the quiet infrastructure of discovery — built not for recognition, but for the researchers who depend on it every single day.