I am a French AI student and engineer. I am finishing two degrees in parallel: an MSc in Engineering at CentraleSupélec, a leading French engineering school (Grande École), and a research master (M2) in Mathematics & Artificial Intelligence at Université Paris-Saclay. Alongside my studies, I work as a Generative AI Engineer at MBDA.

My research interests are large language models, reinforcement learning (offline and multi-agent), world models, and diffusion models / stochastic differential equations. I like implementing research papers from scratch in PyTorch to understand them in depth, and I am working toward a research career at the frontier of these topics, with the goal of doing a PhD.

I am looking for a PhD or research internship in Europe, starting in autumn 2026, on diffusion models and world models. If you work on these topics and think there could be a fit, feel free to reach out (LinkedIn or email below).

A selection of my projects, mostly from-scratch implementations and reproductions of research papers:

• Decision-Transformer-LOB-Trading : offline reinforcement learning via sequence modeling (Decision Transformers), applied to limit order book data and framing high-frequency trading as conditional sequence modeling.
• proximal-diffusion-models-pytorch : Proximal Diffusion Models implemented from scratch (reverse-time SDE through a proximal operator), reducing the number of sampling steps compared to score-based diffusion.
• pricing_collusion : reproduction of Calvano et al. (2020) in a custom Gym environment, showing emergent algorithmic collusion through Q-learning (multi-agent RL).
• ADVI-Taxi-Trajectorie : Automatic Differentiation Variational Inference scaled to 1.7M taxi trajectories (non-conjugate Gaussian Mixture Model, ELBO maximization).
• NSA_Malsiner_2016 : reproduction of "Model-based clustering based on sparse finite Gaussian mixtures" (Malsiner-Walli, Frühwirth-Schnatter & Grün).
• Projet_TDL : reproduction of "(S)GD over Diagonal Linear Networks: Implicit Bias, Large Stepsizes and Edge of Stability".

• Generative AI Engineer, MBDA (apprenticeship, Sep 2025 - present) : research on multi-agent LLM systems trained with reinforcement learning, and self-hosted LLM serving with vLLM and Docker on a GPU cluster.
• Data Scientist, TotalEnergies Digital Factory (apprenticeship, Nov 2023 - Sep 2025) : machine learning for EV-charging, including a transparent dynamic-pricing model and a station-accessibility predictor.
• Member, Automatants, the CentraleSupélec AI student association (Oct 2024 - Nov 2025) : technical workshops on neural networks, CNNs, GANs, Transformers and reinforcement learning.
• President, Led a student association providing residential Internet access; maintained two network infrastructures.

• M2 Mathematics & Artificial Intelligence (research master), Université Paris-Saclay, 2025 - 2026
• MSc in Engineering, CentraleSupélec, major in Data Science, 2023 - 2026
• Exchange in AI for Engineering, Beihang University, Beijing, 2025
• Preparatory classes (CPGE), La Martinière Monplaisir, 2021 - 2023

Languages & tools: Python, PyTorch, vLLM, Docker, Git, GPU compute clusters. I also run a personal RTX 5090 workstation to fine-tune and benchmark open models.

• LinkedIn: in/clement-callaert
• Email: clement.callaert@student-cs.fr

♟️ Outside research, I play chess (around 1750 on chess.com).