Clodéric Mars

AI Product Engineer & Tech Leader

Montréal, Canada

I specialize in human-AI collaboration, with one hand in leadership and the other in development. The products I built have:

empowered artists to direct autonomous characters in movies,
enabled engineers optimize energy & transportation infrastructures, and
assisted instructors in delivering personalized training simulations.

Key skills: Applied Artificial Intelligence, Technical leadership, Software architecture, Program management, B2B presales & project management, Technology transfer, Distributed systems, Cloud infrastructure, MLOps, Full-stack development, Written and oral communication, French (native), English (full professional proficiency).

Experiences

VP of Engineering - AI Redefined (www.ai-r.com)

since 2021/02 - Montréal, Canada

Directed the technology and research distributed team of 16 engineers and researchers across 3 product teams, generating ~$1.5M ARR, 9 peer-reviewed papers, and 3 patents.
Architected and developed Cogment (cogment.ai), evolving it from a technological proof of concept into a robust, open-core MLOps platform adopted both internally and externally.
Designed and trained decision-making AI agents collaborating with humans and leveraging deep reinforcement learning, large language models and symbolic AI to solve complex real-world problems.
Forged collaborations across industries including aerospace and defense (e.g. Thales (www.thalesgroup.com), CAE (www.cae.com), and NAV CANADA (www.navcanada.ca)) ; energy (e.g. RTE (www.rte-france.com)) ; manufacturing (e.g. Renault (www.renaultgroup.com)) ; and academia, (e.g. the University of Alberta (www.ualberta.ca), MILA (mila.quebec), and IRT SystemX (www.irt-systemx.fr)).

Skills: Reinforcement learning, Deep learning, Large Language Models, Simulation, Python, Go, Kubernetes, AWS, Azure.

Freelance AI engineer - AI Redefined

from 2020/08 to 2021/01 - Paris, France

Starting position at AI Redefined as a freelancer before moving to Canada.

Co-founder & CTO - craft ai (www.craft.ai)

from 2015/03 to 2020/04 - Paris, France

Cofounded and developed the company from 7 to 20+ employees, securing ~9M€ in funding.
Directed the product develoment and research team generating ~1M€ ARR, and publishing 5 peer-reviewed papers.
Architected and developed an explainable AI SaaS for time series modeling & forecasting, enabling trustworthy personalized decision support and automation across industries.
Deployed the product across ~10 AI projects, primarily in the energy sector, including with EDF (www.edfenergy.com) and Total (www.totalenergies.fr), helping operators and end-users optimize their consumption.

Skills: Explainable AI, Machine learning, Data science, Python, Rust, Javascript, Kubernetes, Kafka, AWS, Azure.

Lead Software Engineer & Product Manager - MASA Group (masasim.com)

from 2012/01 to 2015/02 - Paris, France

Led a team of ~5 software engineers across Paris and Munich developing a behavior tree and spatial reasoning engine.
Architected and developed an AI agent authoring tool and runtime empowering instructional and game designers to build AI agents for simulations and video games, creating key intellectual property that later became the foundation of craft ai.

Skills: Game AI, Computational geometry, Simulation, Qt, C++.

R&D Software Engineer - Golaem (acquired by Autodesk) (golaem.com)

from 2010/01 to 2011/12 - Rennes, France

Took ownership of the navigation engine for virtual crowds, industrializing research prototypes and algorithms developed at Inria.
Applied crowd simulation technology to diverse use cases, including train car evaluation for SNCF (www.groupe-sncf.com), real-time virtual city population with Dassault Systèmes (www.3ds.com) and crowd simulation for a historical documentary with Mikros (www.technicolor.com/studios/mikros-animation/), which contributed to developing the leading crowd simulation product for the animation and VFX market.

Skills: Game AI, Computational geometry, Simulation, Qt, C++.

R&D Software Engineer - Inria (www.inria.fr)

from 2009/02 to 2009/12 - Rennes, France

Starting position at Golaem as a member of the Bunraku (www.irisa.fr/bunraku) team.

Software Engineer - Teamlog (acquired by Open) (www.open.global)

from 2006/07 to 2009/02 - Rennes, France

Developed a toolchain to create virtual cities for Orange (www.orange.fr), resulting in pedestrian-level digital recreation of Paris and Barcelona.
Developed a contribution system for PagesJaunes.fr (www.pagesjaunes.fr), enhancing the platform’s user engagement.

Skills: Geographic information systems (GIS), Computational geometry, Simulation, Qt, C++, Java.

Additional activities

Contributor - The Good AI

from 2020/07 to 2020/12

Contributed written content about responsible AI.

Data Scientist (volunteer) - APHP (www.aphp.fr)

from 2020/05 to 2020/06 - Paris, France

Supported early COVID-19 epidemiological studies by extracting & preparing data.

Skills: Data science, Python, Apache Spark.

Track chair - nucl.ai

from 2015/01 to 2016/07 - Vienna, Austria

Curated content, managed speakers and presented tracks at a yearly AI developer conference.

Education

MEng (Diplôme d'ingénieur) in Computer Science - INSA de Rennes (www.insa-rennes.fr)

from 2001/09 to 2006/06 - Rennes, France

Engineering program, with a specialization in computer science.

MSc in Artificial Intelligence & Computer Graphics - Université de Rennes 1 (www.univ-rennes.fr)

from 2005/09 to 2006/06 - Rennes, France

Academic curriculum focused on Artificial Intelligence, including machine learning and symbolic reasoning, & Computer Graphics, including virtual reality and animation.

Member of Inria (www.inria.fr)'s Bunraku (www.irisa.fr/bunraku) team, advised by Fabrice Lamarche.

MOOCS

2020/11Coursera & deeplearning.ai - Build Better Generative Adversarial Networks (GANs)Deep Learning
2020/11Coursera & deeplearning.ai - Build Basic Generative Adversarial Networks (GANs)Deep Learning
2020/06OpenClassrooms & Groupe INSA - Utilisez l'apprentissage par renforcement avec un droneReinforcement Learning
2020/05OpenClassrooms & CentraleSupelec - Réalisez des calculs distribués sur des données massivesData Engineering, Distributed Computing
2011/12Stanford University & ml-class.org - Machine Learning courseMachine Learning

Peer-reviewed publications

Full list on Google Scholar (scholar.google.com/citations?user=fvCC1rwAAAAJ)

2024/06 - AASG workshop (Autonomous Agents for Social Good) at AAMAS, Auckland, New ZealandTowards Human-Steerable Battery Energy Storage System Optimization: A Novel MDP FrameworkSai Krishna Gottipati, Clodéric Mars, Julien Gabaud, Vahid Abdollahi, Laila El Moujthahid, Matthew E. Taylor
2023/11 - I/ITSEC (Interservice/Industry Training, Simulation and Education Conference), Orlando, FloridaImmersive AI assistance during eVTOL multi-agent ATC traffic routingJean-François Delisle, Simon Riendeau, Clodéric Mars, Sagar Kurandwad
2023/05 - AAMAS (International Conference on Autonomous Agents and Multiagent Systems), London, UKHiking up that HILL with Cogment-Verse: Train & Operate Multi-agent Systems Learning from HumansSai Krishna Gottipati, Luong-Ha Nguyen, Clodéric Mars, Matthew E. Taylor
2023/05 - ALA (Adaptive and Learning Agent) @ AAMAS (International Conference on Autonomous Agents and Multiagent Systems), London, UKWIP: Human-AI interactions in real-world complex environments using a comprehensive reinforcement learning frameworkMd Saiful Islam, Srijita Das, Sai Krishna Gottipati, William Duguay, Clodéric Mars, Jalal Arabneydi, Antoine Fagette, Matthew Guzdial, Matthew E. Taylor
2020/04 - EGC (Extraction et Gestion des Connaissances), Brussel, BelgiumThe three stages of Explainable AI: How explainability facilitates real world deployment of AIClodéric Mars, Rémi Dès, Matthieu Boussard
2017/07 - CNIA (Conférence Nationale sur les Applications Pratiques de l'Intelligence Artificielle), Caen, FrancePeriodic split method: learning more readable decision trees for human activitiesMatthieu Boussard, Clodéric Mars, Rémi Dès, Caroline Chopinaud
2015/06 - Game AI Pro 2Hierarchical Architecture for Group Navigation BehaviorsClodéric Mars, Jérémy Chanut

Talks

2023/10/25Building AI Assistants for Augmenting Human Operators: A Roadmap Towards Creating SuperteamsAI DevWorld, Santa Clara, California
2023/02/15Trustworthy Human-Machine Teaming: The Next Stage of EvolutionREAIM (Responsible AI in the Military Domain), The Hague, Netherlands
2022/03/23Towards Human-AI Teaming: Intelligence Ecosystems to Tackle High-Stakes Use CasesData Council, Austin, Texas
2017/03/09Améliorer l'expérience utilisateur grâce à l'intelligence artificielle"Hello! Business Talks" Humanité(s) Digitale(s), Paris, France
2014/03/17Environmentally Conscious AI: Improving Spatial Analysis and ReasoningGDC (Game Developers Conference) AI Summit, San Francisco, California
2011/06/24Towards Directable Autonomous CrowdsParis Game/AI Conference, Paris, France

Other Publications

2021/06The future of human-AI systems is already here – it’s just not evenly architectedMilitary Embedded Systems
2020/11AI & Biases 101: The bakery caseThe Good AI
2015/06BT 101 – Behavior Trees grammar basicscraft ai's blog