Home

Welcome to my webpage !

In a nutshell: I am a freelance expert in Artificial Intelligence (AI), especially in Computer Vision topics. I can supervise an AI project from presales, framing, data collection and labeling, to modeling and deployement.

I believe that early discussions with end-users of AI-models or products and subject matter experts is key to launch and run successful projects. This belief made me become a great communicant to any profile, even non-technical.

In addition to that, thanks to my academic background, if you wish to start an AI Research team, I can help you define a roadmap to maximize its impact at a fast pace.  So if you wish to collaborate with me, please reach out !

My background: I have strong applied Mathematics skills and I was specialized in machine learning and image processing during my MVA Master at ENS Paris Saclay in 2012. After my PhD at Télécom Paris in satellite image processing and a few years working in academia, I decided to apply my expertise and rigorous scientific methodology to the private sector. So I joined Fieldbox and worked there for 5 years. It’s a company that increases the operational efficiency of industries with data driven and AI solutions. I was a lead data scientist on client projects and a great help to the presales team. I also co-founded an AI R&D team. It grew so nicely that I became Head of AI Research, managed the team and supervised the teammates on the applied research questions I defined and prioritized.

My main research interests in random order are:

    • Image processing and computer vision (denoising, classification, segmentation, feature matching, object detection, OCR, depth estimation…)
    • Patch-based dictionary learning
    • Gaussian mixture models
    • Signal processing on graphs with variational approaches
    • Predictive models for time series and tabular data
    • Deep learning: Variational Auto-encoders for industiral data, Self Supervised Learning, Vision Transformers, Interpretability
    • Data: SAR images, natural images, 3D point clouds, time series, tabular data…
    • Recently, I started to upskill my game in LLMs and multimodal models