Academic
This page is here for one reason. Anyone can learn to prompt well in a month. Far fewer people can tell you why a retrieval system quietly stops working, or what a benchmark number is hiding. That is the part I was trained for.
Below are the degrees, the paper, the thesis and the research projects. Each one carries the part that does not flatter it, because I would rather you found the caveat here than worked it out later.
Université Paris-Saclay · Paris, France
Machine learning, deep learning, information theory, information retrieval and data engineering, taught in English and French. M1 completed June 2026 with an overall mark of 15.097/20; M2 runs to June 2027.
2025 – 2027
Charles University, Prague · Prague, Czechia
Algorithms, software engineering and mathematical logic, specialising in machine learning. Thesis on procedural terrain generation under fixed-point constraints, defended September 2025 and later extended into the IEEE CoG 2026 paper. Won two public-speaking competitions.
2021 – 2025
University of Cape Town · Cape Town, South Africa
A year of business science (statistics, economics and finance) before moving to Prague for computer science. It is why the commercial side of this work is not something I had to learn later.
2020 – 2021
I also taught. I ran lab sessions and marked coursework for undergraduate computer science at Charles from September 2024 to June 2025. Explaining one idea to twenty people who each misunderstood it differently is the nearest thing to consulting practice I have had.
Each one with the reason to discount it
Petr Šimůnek, Oliver Wakeford
IEEE Conference on Games 2026 (CoG), Madrid
Accepted for oral presentation, full paper. 89 of 239 papers (37%)
Second author. I built the terrain tool and proposed the constraint-propagation idea in my BSc thesis; my supervisor extended it into the paper, ran the experiments and led the writing. Reviewers noted the evaluation is limited to a single authored benchmark.
Oliver Wakeford
BSc thesis, Charles University, Faculty of Mathematics and Physics
Defended 5 September 2025
Oliver Wakeford
Master's internship report, ACL format
Submitted July 2026. Not peer-reviewed, and not published
An internship deliverable assessed by the university, not a publication. The corpus belongs to a client and cannot be shared.
The conference figure, since it is the one that always gets quoted: 89 of 239 papers (37%). That is a real bar, and it is not a top-tier venue. The paper's headline improvement also sits within about 0.4% of inverse distance weighting, which is the simplest sensible alternative.
Most of these are group projects and each one says so, with the collaborators named. The exceptions are the interactive-ML app, which is solo, and the retrieval study, which is mine but rests on a client corpus I cannot share.
The most useful thing in that list is not a score. It is the retrieval study, where rerunning my own analysis turned up six defects in my pipeline, four of which had already put a wrong number in a document, and the report retracts the claim rather than quietly restating it.
Read what went wrong