Technology leadership

Harvard School of Engineering and Applied Sciences: an IT, research, and academic computing turnaround

As the inaugural CIO and Assistant Dean for Computing at Harvard's School of Engineering and Applied Sciences, Gabriele Fariello led all of the school's computing from day one: information technology, research computing, and academic computing. He rebuilt a fractured organization, cut real IT spend by nearly 30% over three years, and, in the same period, expanded the research computing available to faculty and students roughly fifty-fold while user satisfaction rose.

The situation

In 2013, Harvard SEAS created its first CIO and Assistant Dean for Computing role and asked Gabriele to take it on. The mandate was difficult. He had to repair a fractured IT and computing organization, mend a strained relationship with the wider Faculty of Arts and Sciences over research computing, and bring rising IT costs under control. All of this had to happen while the computing that faculty and students depended on kept running and improving.

What he did

Gabriele reorganized the operating model rather than defending an expensive, siloed status quo. He moved commodity services to shared university and third-party providers, stood up a lean team focused on the school's genuinely idiosyncratic needs, and reinvested the savings into a modern, high-value computing capability. The centerpiece was consolidating the school's high-performance computing into Harvard's shared research-computing environment.

That consolidation required persuading faculty to give up some local control in exchange for far greater capability. Gabriele earned that agreement with a practical, low-risk approach: try the shared environment, and if it fell short he would fix it. The results made the case.

The results

• Up to ~$13.9M in cumulative savings and avoided cost (FY13-FY16), corroborated by the school's finance office.
• The fifty-fold expansion came from consolidating into Harvard FAS Research Computing's Odyssey environment (publicly documented at more than 60,000 CPUs and 15 PB of storage by 2015).
• Per-capita IT spend down 31 to 44%, at a time when no peer Harvard organization achieved reductions greater than 10%.
• Data-center footprint reduced while reliability and capability rose.
• Savings came from operating discipline, not service cuts: moving faculty from under-desk workstations and private hardware to centrally managed systems, a standardized purchasing program with bulk end-of-cycle buys, self-insurance in place of extended warranties, pre-imaged ready-to-use machines, and a standing rule that no renewal was paid without renegotiation (which alone cut contract costs by about 16%).

The cloud decision

Gabriele took SEAS fully to the cloud, the first school at Harvard to do so. He announced it at the University CIO Council two weeks after a major AWS outage in 2013, when moving to the cloud was still contested and colleagues thought the timing was reckless. His argument was about accountability and economics rather than fashion. The school's own data centers ran below 99% uptime, could not justify around-the-clock coverage, and left SEAS fully liable when they failed. A hyperscale provider offered reliability and durability guarantees the school could never match on its own. He also worked with Harvard's University Chief Information Security Officer to develop the cloud-security standards that other schools later adopted. Roughly 85% of critical infrastructure moved to the cloud, estimated uptime rose from about 98.6% to about 99.8%, and direct costs fell by more than one hundred thousand dollars a year. The larger gain was leverage: the move let Gabriele redirect roughly three and a half full-time staff from commodity infrastructure to research and academic computing, reaching total-cost parity with far more value delivered for the same spend.

The leadership lesson

The SEAS turnaround is a value-creation story. Spend fell while capability, trust, and research computing all expanded, which is what disciplined operating leadership produces under pressure. Gabriele brings that same pattern to institutions where the budget and the mission are both under strain: he expands what the organization can do and restores confidence in it, and the economics improve as a consequence.

Relevance to regulated AI and data mandates

The capability on display here transfers directly to the primary spike. A regulated-industry CAIO, CIO, or AI-transformation executive has to expand what an organization can do with data and AI, win the trust of skeptical technical and scientific stakeholders, and hold financial discipline at the same time. This case shows all three at institutional scale, and it shows scale through the complexity of consolidating high-performance research computing rather than through raw headcount. Boards that need capability and financial discipline together are looking at the same skill set.

SmartNode case study  ·  Harvard neuroinformatics case study