Freelancer profile translated to English.

Description

I develop **artificial intelligences capable of running where others cannot**.

Microcontrollers, constrained environments, real-time processing, industrial vision: my work consists of **designing and optimizing ultra-efficient neural networks**, capable of executing models with **a few hundred kilobytes of memory and less than a milliwatt of power consumption**.

I design my solutions as a **bridge between research and real-world application**: each AI I develop must strike the right balance between **scientific performance, technical efficiency, and business value**.

My approach: combine **research rigor with industrial requirements**, to transform artificial intelligence into **useful, integrable, and sustainable technology**.

From embedded vision to language processing, including data analysis, I help companies **bring effective, elegant, and controlled AI to life**.

Industry field of expertise

Languages

French
Native or bilingual
English
Fluent

Workplace preferences

Remote only

Primarily works remotely

Micelia
Co-founder & Chief Research Officer (CRO)
SOFTWARE PUBLISHING
July 2025 - Today (11 months)
Grenoble, France
I co-founded the companyMiceliato help businesses develop custom artificial intelligence technologies, fromapplied researchtoindustrializationand **skills transfer**.

Within the company, my role is to ensure thescientific and technical coherenceof R&D work, guaranteeing that research choices and technological directions meet our clients' concrete needs. I also work on projects with a stronginnovationcomponent:

*Real-time embedded AI**, my specialty: design and optimization of ultra-compact neural networks for execution on microcontrollers and specialized processors, in environments with **extreme resource constraints(a few hundred KB of RAM, sub-mW consumption).Low-levelsoftware optimization (C, ASM) and hardware/software co-design to maximize performance while minimizing latency and energy consumption.

*Computer vision(classification, detection, segmentation, tracking) applied to health, industry, and the environment.

*Data analysis and machine learningon tabular data for modeling, prediction, and decision support.

*Natural Language Processing (NLP)for text analysis, content generation, and semantic understanding.

In parallel to myfreelanceactivity, I rely onMiceliato leverage complementary expertise (Cloud, LLM, MLOps). This network allows me to ensuretechnical excellenceacross the entire AI scope and to be asingle point of contactfor all project stages—from research to production.
Computer Vision Embedded Systems Deep Learning Data Science NLP
Kalray
Embedded AI/Software Engineer
TECH
March 2023 - June 2025 (2 years and 3 months)
Montbonnot-Saint-Martin, France
R&D and optimization of embedded artificial intelligences.

*Key contributionto Kalray's ONNX graph compiler: translation of ONNX graphs to an internal representation and implementation of advanced graph and execution planning optimizations.

*Manager of the “Libtensors” library: development andoptimizationof high-performancecompute kernelsin **ASM**, **C**, and **C++**, grouped into a shared library used in production.

*AI Architect:co-definition of specificationsandAIrequirements for Kalray's next-generation accelerators, with a focus on **performance**, **scalability**, and energy **efficiency**.

Experiencecombining low-level R&D, software optimization, and embedded AI architecturein a high-performance environment focused on massively parallel computing.
Embedded Systems C/C++ Programming Performance Optimization Real-time Processing Industrial Vision
Neovision
R&D embedded AI engineer
SOFTWARE PUBLISHING
March 2022 - March 2023 (1 year)
Grenoble, France
Study, design, and production deployment of AI solutions optimized for microcontroller execution.

Development of compact and high-performance neural networks adapted to the constraints of embedded computing:

*Design and trainingof lightweight models for recognition and detection on resource-constrained systems ( **< 1 MB RAM**, **< 1 mW consumption**).

*Optimization and quantizationof networks for real-time deployment on microcontrollers ( **Edge AI**).

*Complete integrationinto an embedded production chain, from research to hardware validation.

Experience at the intersection of **embedded artificial intelligence**, **neural model optimization**, and **systems engineering**, with an approach focused on **performance**, **reliability**, and **very low energy consumption**.
Deep Learning Embedded Systems Real-time Processing Edge AI Computer Vision