Célia Benquet

Célia Benquet

Doctoral Researcher

I’m a fourth-year doctorate researcher in computational neuroscience at EPFL’s M-Lab of Adaptive Intelligence. I develop machine learning methods for biological time series, and my experience includes neural recordings and wearable cardiac signals. I’m interested in applying AI to problems in medtech or biotech.

Interests
  • Computational neuroscience
  • Representation learning
  • Time-series
  • Vision and decision-making

Education

  • PhD in Electrical Engineering, 2027

    PhD Excellence Fellow EPFL, Switzerland

  • MSc in Computational Science & Life Science Engineering, 2022

    Thesis at Harvard University, US (EPFL-WISH Fellow) EPFL, Switzerland

  • BSc in Life Science Engineering, 2019

    Exchange at KTH, Sweden Thesis at Cambridge University, UK EPFL, Switzerland

Experience

 
 
 
 
 
EPFL, MLAI, Prof. Mackenzie Weygandt Mathis
Doctoral Researcher
May 2023 – May 2027 Geneva, Switzerland
Developed Unified CEBRA to compare neural recordings across experiments. Led development of an augmented reality platform across four labs (20 authors) and mentored six students and research assistants. Related publications
 
 
 
 
 
EPFL, MLAI
Research Software Engineer
Oct 2022 – May 2023 Geneva, Switzerland
Contributed model architectures, data and inference pipelines, and API improvements to the open-source CEBRA library. CEBRA on GitHub
 
 
 
 
 
Harvard University, Uchida laboratory, Prof. Naoshige Uchida
Visiting Graduate Student
Feb 2022 – Sep 2022 Cambridge, US
Studied how neural populations and recurrent models represent beliefs during decision-making. Related preprint
 
 
 
 
 
Rhythm Diagnostic Systems (medtech startup)
R&D ML Engineer Intern
Aug 2021 – Feb 2022 Strasbourg, France
Built reproducible evaluation pipelines and benchmarked anomaly detection methods on clinical and synthetic ECG and PPG data for a wearable cardiac monitor.

Open-Source Contributions

  • CEBRA (1.1k GitHub stars) — Representation learning for neural and behavioral data.
  • FreelyMovingVR4Mice — Virtual reality tools for experiments with freely moving mice.
  • CEBRA-lens — A Python library for interpreting CEBRA models, in development.

Publications

First-author publications

Visual uncertainty and task demands shape active sensing strategies in mice
Mice flexibly adjust active sensing strategies in response to uncertainty and task demands.
Visual uncertainty and task demands shape active sensing strategies in mice

Other publications

Dynamic belief representation and updating through learned attractor-like dynamics in the frontal cortex
In mice, orbitofrontal cortex activity develops task-specific, attractor-like dynamics that track changes in inferred beliefs.

Skills

Machine Learning

Representation learning and time-series modeling with PyTorch and scikit-learn

Data Analysis

Neural and physiological signals, behavioral data, and model evaluation

Software Engineering

Python, C++, Docker, CI/CD, and reproducible APIs

Talks & Posters

Talks

Posters

  • Nov 2026 — 2026 SCENE annual meeting. Unified CEBRA learns shared neural latents across recordings to map single-neuron contributions to population codes.
  • Feb 2025 — Swiss Society for Neuroscience Annual Meeting 2025. An augmented reality system to study active visual sensing in dynamically occluded environments.
  • July 2024 - Neuro-X Retreat 2024. An augmented reality system to study active visual sensing in dynamically occluded environments.

Teaching

Teaching assistant at EPFL for System Neurosciences, Basic Neurosciences, Basic C++ and Algorithms, Complex and Vectorial Analysis, and Thermodynamics and Relativity.

Interests

Environment

Facilitator at La Fresque du Climat

Running

Marathon, semi-marathon, trails

Mountain sports

Hiking, skiing, climbing

Acting

Plays, now trying improv

Water sports

Sailing, kite-surfing, scuba-diving

Reading

Non-fiction and fiction