Description

Ingénieur expérimenté qui peut apporter une maturité technique aux applications intensives en calcul et en données.

Systèmes informatiques:

Passionné de C++. Compétent en C et en Python. Prêt à apprendre rapidement de nouveaux langages et paradigmes comme Go, Scala et JavaScript.
Ingénierie de la performance logicielle et programmation parallèle (CUDA, OpenMP).
Systèmes de traitement des données et d’apprentissage automatique: SQL, NoSQL, Pandas, MapReduce, PyTorch.
Principes de conception de logiciels.
Outils Linux. Systèmes de construction logicielle. Docker.

Apprentissage automatique appliqué et mathématiques appliquées:

En particulier, le traitement de l’image et la vision par ordinateur. Autres expériences: robotique, publicités en ligne, séries chronologiques et textes.
Bonne compréhension des concepts d’apprentissage automatique (y compris l’apprentissage profond).

Expérience des environnements de production et de recherche.

Français: B1 (Intermédiare).

Languages

Workplace preferences

Can work on-site

Paris (up to 50km)

PacketAI
Machine Learning Engineer
September 2020 - March 2022 (1 year and 6 months)
Paris, France
Implemented the root cause analysis system and parts of the metric anomaly detection system to monitor IT infrastructure: Python, Cloud technologies.
Clarified the behavior of a broad range of ML algorithms during discussions. Discovered new leads in literature.
TheraPanacea
Research And Development Engineer
February 2020 - September 2020 (7 months)
Paris, France
Speed-up of the therapy planning system. Proposed and implemented (in Python) approaches to compute the dose influence matrix faster based on low rank and multiscale representations.
INRIA
Postdoctoral Researcher
July 2018 - January 2020 (1 year and 6 months)
91400 Saclay, France
Part of the DataShape team. Optimization problems based on topological features:
- Theory and implementation of stochastic gradient descent for optimization based machine learning problems involving topological features. Co-author of "Optimizing persistent homology based functions", ICML 2021. Selected for long oral presentation (top 3%).
- Demonstrated the benefit of multiscale representation to reach a better local minimum for optimization based on topological features.
- Comparing time series using topological features. Work featured in Fujitsu press release: https://www.fujitsu.com/global/about/resources/news/press releases/2021/0716-01 .html Implementation in Python, PyTorch and C++.