# ausset / me


Last edited: June 1, 2026


I'm a machine-learning researcher and engineer with a PhD in survival analysis. I work on calibrated risk models, model validation, and production ML systems for regulated, high-stakes decisions. Recent work includes credit scoring, model monitoring, document automation with vision-language models, and retraining/backtesting pipelines. I'm interested in AI evaluation, health AI, model quality, and finance/quant ML.

## CV / contact

You can find my full CV [\[here\]](/cv.pdf){aria-label="full CV PDF"} or [\[here\]](/cv_compact.pdf){aria-label="one-page CV PDF"} if you prefer one-pagers.

You can write to me at `guillaume at ausset dot me`. You can verify I am indeed me by using [`PGP`](/~aussetg.asc).
If you are a recruiter, I will definitely read your mail if it is `PGP`-signed.

## Selected experience

I am currently taking time off before my next professional endeavor to study things I could not justify doing professionally: applications of AI to hyper-personalized medicine, experiments in agentic AI, learning to build, understand, and fly multirotor UAVs (i.e., drones), and a myriad of other riveting subjects, from late interaction retrieval to quantization.

Before that, I worked for Younited Credit, where I started by reviewing all existing ML models as part of a team of two for the newly created Model Risk Management team. After that large effort reviewing the backlog and issuing a number of recommendations to the Data Science team to improve their processes, I joined the Data Science team to help apply the state of the art in default prediction to improve credit decisions and put the recommendations into action.
I ended my tenure at Younited Credit in the restructured Data Science team, now separated from the Credit Scoring team, where I took ownership of the lifetime value prediction model that powered the ads bidding process, as well as the model responsible for automating the extraction of structured income data from various Italian documents.

I worked with Télécom Paris and the CDC on bias in machine learning, where we tried to understand, analyze, and eliminate the various forms of bias that people looking for state-funded training can face.

During my PhD, I was fortunate to work on applications to medicine, and I am currently working on applications in healthcare (though not as much as the *other [Ausset](https://scholar.google.fr/citations?user=LuGvY18AAAAJ&hl=fr)*).

In another life, I studied stochastic calculus and mathematical finance and worked as a quant. You can still find cheat sheets I wrote for exams [here](/cheatsheets){aria-label="cheat sheets for exams"}.

## Current interests

I'm currently very interested in N-of-1 trials and Bayesian statistics, ColBERT-style late interaction retrieval, and drones.
I love [Julia](https://julialang.org/) and think more people should give it a try.