Paul Chambaz

Hey! I’m finishing my Master’s in Computer Science (AI2D track) at Sorbonne Université. I’m interested in reinforcement learning because it can achieve incredible performance in real-world tasks, yet we still don’t understand why many algorithms work or fail. I’m currently at ISIR with Prof. Olivier Sigaud, working on a new RL algorithm we aim to submit to ICLR. I believe in transparent, controllable AI rather than closed black boxes. Outside of my studies, I’m reading, running or writing open source software, including tools for ALIAS, the CS student association at Sorbonne, or a self-hosted audiobook stack.

CV Email Github Scholar LinkedIn

Research

Actor Free critic Updates for Off-Policy and Offline Learning

Paul Chambaz, Frédéric Li Combeau

M1 AI2D Sorbonne Université

Paper Code

Blog posts

Some notes on the TQC figure (Aug 2025)

Estimation biases represent a persistent challenge in reinforcement learning, where errors in value estimation can accumulate through bootstrapping and compromise learning efficiency. Among these …

Projects

Polybase - student paper handout distribution system

ALIAS Student Association

go, templ, htmx, tailwindcss, sqlite3

Website Source code

Ssherd - RL training job orchestrator over SSH

paulchambaz

go, templ, htmx, websocket, ssh, nfs, slurm

Source code

Mpcube - album focused terminal music client

paulchambaz

go, bubbletea, mpd

Source code

Iliad - self-hosted audiobook server

paulchambaz

rust, rest, sqlite3

Source code

Odyssey - android audiobook client for Iliad

paulchambaz

kotlin, android

Source code

Odyssey TUI - terminal audiobook client for Iliad

paulchambaz

go, bubbletea

Source code

Education

Master in Computer Science - AI2D

Sorbonne Université (2024-2026)

M2 S3: 15.8/20 1st/45, M1: 16.4/20 1st/53, S1: 15.85/20, S2: 17.03/20

Excellence diploma program

Bachelor in Computer Science

Université Paris Cité (2019-2023)

Mention Très Bien - 16/20

First year bachelor Computer Science

Université Claude Bernard Lyon 1 (2018-2019)

Work Experience

M2 Research Intern

ISIR Laboratory, Sorbonne Université (February - August 2026)

Supervised by Prof. Olivier Sigaud

Python, JAX, Pytorch, Reinforcement Learning, Matplotlib

M1 Research Intern

ISIR Laboratory, Sorbonne Université (Summer 2025)

Supervised by Prof. Olivier Sigaud

Python, JAX, Pytorch, Reinforcement Learning, Matplotlib

Cybersecurity Developer

Mobeta (February - August 2024)

Supervised by Arthur Le Corguillé

TypeScript, Go, Python, Docker, Cybersecurity

OSINT Developer Intern

Lexfo (Summer 2023)

Supervised by Armand Sylvain

Python, Ansible, Active Directory, Proxmox

Coursework

M2 Sorbonne Université (S3)

UM5IN253 - Models and Algorithms for Decision under Uncertainty

UM5IN256 - Models and Algorithms for Multicriteria and Collective Decision

UM5IN257 - Algorithms for Optimization and Game Theory

UM5IN259 - Artificial Intelligence and Robotics

UM5IN861 - Deep Learning

M1 Sorbonne Université (S2)

MU4IN204 - Decision and Games

MU4IN201 - Problem Solving

MU4IN202 - Foundations of Multi-agent Systems

MU4IN811 - Machine Learning

MU4IN206 - AI2D Research and Development Project

MU4IN207 - Learning and Robotics

M1 Sorbonne Université (S1)

MU4IN800 - Logic and Knowledge Representations

MU4IN601 - Probabilistic and Statistical Methods and Algorithms for Computer Science

MU4IN200 - Modeling, Optimization, Graphs, and Linear Programming

MU4IN600 - Basics of Image Processing

MU4IN900 - Complexity, Randomized and Approximate Algorithms

MU4IN400 - Concurrent and Distributed System Programming