SECKIN OZBEK Seçkin Özbek
I am a data scientist and a former career diplomat. Eight years at the Ministry of Foreign Affairs of Türkiye, including a posting in Geneva in multilateral economic diplomacy at the Permanent Mission to the WTO, and an MSc in Applied Social Data Science at the London School of Economics, in the Department of Methodology, as a Chevening Scholar. My work sits where quantitative method meets institutions: causal inference, machine learning on political and legal text, and the measurement of things that resist measurement. I contribute quantitative measurement to the Values for Cohesion project at the University of Birmingham.
I build verification systems rather than generative ones. Project Shimmer reviews regulated documents under a written constitution, where no model may audit output from its own model family. SILICA-Bench tests whether a frontier model catches the silent failures in a research design that a trained methodologist would catch: a weak instrument, an absent diagnostic, a method asserted outside its regime. The methods piece behind this work, "Who Checks the Checker", is on Medium. More on the projects page.
Eight years at the Ministry of Foreign Affairs of Türkiye. First Ankara, on bilateral economic and defence relations with the United States and Canada. Then three years in Erbil, running the visa and consular sections of the Consulate General. Then Geneva, at the Permanent Mission to the WTO, representing Türkiye at UNCTAD and the International Trade Centre, and serving as Rapporteur of the thirteenth session of the UNCTAD Trade and Development Commission. Back in Ankara, I held the country file for the World Economic Forum, the ILO, the FAO and the World Food Programme, and was later the operational focal point for the President's G20 Johannesburg Summit file.
MSc in Applied Social Data Science at the London School of Economics and Political Science, as a Chevening Scholar, following a BA in Economics from Ege University. The dissertation, "Values in UN General Debate Speeches", awarded a distinction, applies machine learning and natural language processing to classify leaders' statements against ten World Values Survey dimensions, and assesses their alignment with public values and with the Secretary-General's own assertions, operationalising the norm life cycle, norm clusters and soft power.
If your organisation faces governments or international organisations it has to engage, owes a funder a measurement plan, produces analysis someone has to stand behind, is deciding where AI belongs in its operations and how to answer for it, or needs the real effect of a programme or a policy measured where randomising is impossible, that is what I take on. What an engagement looks like, and how to start one, is on the collaborate page.