Kofi E. Ampaabeng

About — Tech leader

I’m the “can-do” engineer-leader — hand me the new, undefined problem, especially one in a field nobody’s mapped or with the obvious tools off the table. I learn it, and I ship it.

Walk into unfamiliar, constrained territory, do the real work, leave a system that holds up, and build the team that keeps it running.

I started as an economist. The PhD taught me to reason about measurement — what the data actually lets you claim — and the measurements I needed usually didn’t exist yet. So I learned to build them, then to build the teams that could: at Mercatus I directed the Quantitative Health Lab and the Open Health Policy program and shipped HealthRegData on ML models we trained ourselves; I architected RegAuth to render the full U.S. federal legal corpus as structured data — years before LLMs could do it. Economist to AI leader, one problem at a time.

Since then I’ve led teams and shipped products across regulated banking, data platforms, and applied ML — a core-banking and regulatory platform live for banks across Ghana and South Africa, an IFRS-first ledger where every posting is immutable, three teams built from scratch. The domains change; the method doesn’t — and underneath every one of them, a problem-solver who learns.

20+
Years shipping
3
Teams built from scratch
4
Client countries

01   Why I build

I build products I want to use. I’m the first user of everything here, and most of it started as my own problem.

Being the first audience is the whole method. It’s why I don’t ship half-measures — I have to live with them tomorrow. It’s why governance and getting it right come first — because I care, not because someone filed a ticket. The range is the point: a curious person solving his own problems to a standard he’d accept.

02   How I work

Governance is the architecture, not the paperwork.

ADRs gate decisions before they harden into folk memory; every change carries a ticket and a test; audit trails are append-only. I treat engineering standards the way a bank treats controls.

AI inside controls it can’t bypass.

The model writes the code and opens the merge request; a compliance engine blocks the merge on any violation. Speed without surrendering control — the same discipline a regulated transaction gets.

Architect to the numbers.

Cloud-native or self-hosted, the budget and the business indicators decide — not the trend.

03   Career

  1. Founder & Principal Engineer / Architect

    2024 – Present

    Cape Wesley

    Regulated banking, IAM, ML & mobile portfolio. Hired and built a globally distributed team (Ghana, Rwanda, US) serving clients across South Africa, Ghana, Pakistan, and the US.

  2. Senior Backend Engineer

    2022 – Present

    ECS — USPS Change-of-Address

    Technical owner of a national-scale federal system (cleared). Led a WebSphere→WildFly migration with no prior experience in either; built a custom queue when Kafka/ActiveMQ were barred; re-architected auth to a BFF model.

  3. Co-Founder & Executive Director

    2019 – 2024

    Excenit (fintech startup)

    Built the RSS regulatory & core-banking platform and a banking client base across Ghana and South Africa from zero. Subsumed into Cape Wesley.

  4. Director, Open Health Policy

    2021 – 2023

    Mercatus Center, George Mason University

    Built the Open Health Policy team of research scholars; mentored analysts to published co-authorship.

  5. Director, Quantitative Health Lab

    2020 – 2023

    Mercatus Center, George Mason University

    Led researchers and data scientists applying AI and machine learning to U.S. health-care data; built the HealthRegData product and its data-science team from the ground up, on ML models we trained ourselves.

  6. Senior Research Fellow, Health Policy

    2020 – 2023

    Mercatus Center, George Mason University

    Built the RegCensus data platform and API with Python and R clients; testified before a U.S. Senate subcommittee and two state senates.

  7. Research Fellow and Data Scientist, Policy Analytics Project

    2019 – 2020

    Mercatus Center, George Mason University

    Used machine learning to create data on federal and state regulation and occupational licensing for other researchers.

  8. Project Domain Lead, Health Services Research

    2017 – 2019

    IMPAQ International

    Domain lead to January 2018, then task lead. Across 2013–2019: led project teams for HHS, DOL and USDA; Medicare and Medicaid claims on Teradata for CMS; designed technical-solution architectures for DOL and USDA federal data systems.

  9. Research Associate, Health Services Research

    2016 – 2017

    IMPAQ International

  10. Research Associate, Human Services Research

    2013 – 2016

    IMPAQ International

  11. Banking & Software (Infrastructure Manager, DBA, Sr. Developer)

    2003 – 2007

    Fidelity Bank · Agricultural Development Bank · Bsystems — Ghana

    Oracle infrastructure and core-banking rollouts; led full-SDLC delivery of a financial asset-management system for Ghana's largest insurer.

Résumés on the Résumé page; papers, testimony and commentary on the Research page.

Skills

Counted from the systems on the Work board. Open a row for its parts.

SkillEnterpriseConsumerToolsInfrastructureSystems
Languages and frameworks
623112
21328
40105
12003
Data
523111
40206
20002
ML and LLM
12306
22105
Cloud and DevOps
20237
01225
00325
00145
00112
Operations and security
42208
10225
10124

E Enterprise · C Consumer · T Tools · I Infrastructure

Earlier roles

Teradata · AWS · Data warehousing and ETL · Econometrics · Forecasting