Organized by the work, not a timeline — several of these roles overlapped. Each is one problem, the approach, and what came of it.
AI-Safety Program & Award Operations
Program Director, Future of Life Institute (2020–2025)
Problem
AI-safety research depends on coordination that the field is rarely built to provide: aligning academic and industry partners, recognizing the work that matters, and getting resources to it — all under public scrutiny and across borders.
Approach
I led program strategy across academic and industry partners and ran five cycles of the Future of Life Award, 2020 through 2024. Each cycle is a research process on an annual clock: identifying people whose work averted a catastrophe that then never happened, verifying a claim that by definition left thin evidence through expert interviews and due diligence, and making the case to the board. Two cycles were on nuclear risk — the scientists behind nuclear winter, and the storytellers behind The Day After and WarGames — and the 2024 cycle honored the foundations of computer ethics and AI safety. Delivery ran alongside the research: convenings of 50–150 guests on budgets up to $500K, two internal staff retreats, two international workshops with associated press events, and the operational backbone of travel, financial control, and the projects that kept programs running. I also directed a media-outreach effort on a $50K budget to make the research legible to a general audience.
Outcome
Every cycle ran on budget and across borders without incident, and the evaluation work behind the award informed award decisions above $200K. The outreach effort carried the research to audiences well outside the field, in national and international press.
Learner Engagement Systems
Learner Engagement Manager, Harvard University, John F. Kennedy School of Government (2020–2021)
Problem
Support for 200+ Public Leadership Credential participants was absorbing time that should have gone to the program itself, with no system to see where the load was coming from.
Approach
I ran a stakeholder analysis, implemented a Salesforce CRM to triage and track support, built and maintained the Canvas LMS environments, and used Salesforce and Tableau to analyze enrollment and outcomes. I standardized delivery policies so service was consistent across programs.
Outcome
The system cut the support load 30% — and, more lastingly, converted ad-hoc triage into enrollment and outcome data the program could plan around.
Partner Program Management
Partner Manager, Archibus (2007–2015)
Problem
Reaching the IWMS (Integrated Workplace Management Systems) market meant working through resellers — partners who needed enablement and a reason to prioritize the product.
Approach
I recruited and managed 15 software resellers as their primary point of contact, ran needs assessments and solution recommendations to support their sales cycles, and tracked partner performance to find growth opportunities.
Outcome
The channel grew 30% year over year and stabilized at $12M in annual partner revenue — a book of business that held because the resellers had a reason to keep the product first.
Teaching & Knowledge Communication
Teaching Assistant and Instructor, Harvard Division of Continuing Education (2006–Present)
Problem
Adult learners in history courses need consistent instructional support and someone reliable across the long arc of a program.
Approach
I have provided instructional support for history courses since 2006 — including nuclear deterrence since 2016 — delivering one to two guest lectures per semester, mentoring students and facilitating discussion, managing course logistics, and maintaining the Canvas LMS environments.
Outcome
Continuous since 2006, sustained alongside every program position here rather than instead of them — the reliability shows up over the long arc, not just one project.