Freelancer profile translated to English.

Description

After training in Data Science & Computer Science at École polytechnique and Télécom Paris, I started working in 2017 on language model architectures with attention mechanisms, then I implemented question answering systems for legal and computer vision models for autonomous cars and healthtech.

Co-founder and CTO of Fincome, a B2B SaaS Business Intelligence product, with full responsibility for technical architecture, backend, data systems, infrastructure, and deployment, as well as active contribution to product topics (framing, prioritization, specification). Building the product from scratch, then structuring and managing a technical team of up to 5 people, with a very hands-on involvement in critical topics.

Curious and passionate about science and programming, I am particularly interested in complex systems and their concrete implementation. Today, I am focusing more specifically on fundamental AI issues: representation, reasoning, agency, as well as its applications in science and medicine.

Feel free to contact me to discuss your project!

Industry field of expertise

Languages

French
Native or bilingual
English
Fluent
Spanish
Fluent

Workplace preferences

Remote only

Primarily works remotely

Fincome
Co-founder & CTO
SOFTWARE PUBLISHING
December 2021 - Today (4 years and 6 months)
Paris, France
Full responsibility for technical architecture, backend, data systems, infrastructure, and deployment, as well as active contribution to product topics (framing, prioritization, specification).

Building the product from scratch, then structuring and managing a technical team of up to 5 people, with strong personal and hands-on involvement in key technical topics.
Back-End Development FastAPI Amazon Web Services Terraform DevOps
Groupe Crédit Agricole (AE)
Data Scientist
BANKING AND INSURANCE
August 2021 - April 2022 (8 months)
Montrouge, France
Research & Development, implementation and deployment of AI pipelines (NLP / computer vision) for automatic information extraction from various document types (scanned PDFs, legal contracts, ID documents, etc.).

Topics: document layout analysis and segmentation, optical character recognition (OCR), question answering, information retrieval.

Main technologies: Python, Transformers, OpenCV, PyTorch, Elasticsearch
NLP Python Algorithms Computer Vision OCR
MARELLI SMART ME UP
Data Scientist
AUTOMOBILE
September 2020 - June 2021 (8 months)
Implementation in PyTorch of computer vision scientific papers for autonomous vehicles.

- Learning Joint 2D-3D Representations for Depth Completion (2019)
- ResNeSt: Split-Attention Networks (2020)
- SOLOv2: Dynamic and Fast Instance Segmentation (2020)
- Deep Residual Learning for Image Recognition (2016)
- Panoptic-DeepLab: A Simple, Strong, and Fast Baseline for Bottom-Up Panoptic Segmentation (2019)
- Pseudo-LiDAR++: Accurate Depth for 3D Object Detection in Autonomous Driving (2019)
- AANet: Adaptive Aggregation Network for Efficient Stereo Matching (2020)
- YOLOv3: An Incremental Improvement (2018)
- YOLACT Real-time Instance Segmentation (2019)
- Real-Time High-Performance Semantic Image Segmentation of Urban Street Scenes (2020)
- A deep primal-dual proximal network for image restoration (2020)
Pytorch Python Computer Vision