About
Hi, I'm Jeremie and I live in London. I'm a machine learning engineer at Papercup doing research on large generative models for speech (TTS and voice conversion) to use in dubbing. I also play jazz / gypsy jazz guitar and like rock climbing.
I was previously a statistical scientist at Cervest working on climate modelling. Before that I did a postdoc at Lancaster University in Bayesian machine learning with Chris Nemeth, and a PhD in computational statistics at UCL with Yvo Pokern.
Research:
- Efficient and generalizable tuning strategies for stochastic gradient MCMC [Statistics and Computing, 2023]
- SGMCMCJax: a lightweight JAX library for stochastic gradient Markov chain Monte Carlo algorithms [JOSS, 2022]
- MCMC for a hyperbolic Bayesian inverse problem in traffic flow modelling [Data-Centric Engineering, 2022]
- Ensemble sampler for infinite-dimensional inverse problems. [Statistics and Computing, 2021] [arXiv]
- PhD thesis
Software I wrote / contibuted to:
- SGMCMCJax: lightweight library of stochastic gradient MCMC algorithms written in JAX
- JAX-tqdm: add a tqdm progress bar to JAX scans and loops
- The DjangoVerse: interactive data visualisation of Gypsy Jazz musicians around the world
Music related projects:
- London Django Collective: Collective of a few gypsy jazz players in London. Here’s a video of one of my tunes.
- Gypsy Jazz memes
Other projects:
- Knowitwall: public engagement magazine based at UCL
- The Art of Toastie
Meet:
I’m always happy to chat about science/tech/maths/random! So if you’re in London and want to meet for coffee/tea just email me at jeremie.coullon at gmail.com. I’ll buy the drinks :)