Freelancer profile translated to English.

Description

Machine Learning Engineer with 4 years of experience. Double Master's in Mathematics and Computer Science, specializing in Data Science (2021) from Paris Cité University. Proficient in Computer Vision, MLOps, LLMs. Former Tech Lead at Echo Analytics, with experience in developing deep learning models, ML pipelines, and Cloud infrastructure.

Languages

French
Native or bilingual
Vietnamese
Native or bilingual
English
Fluent

Workplace preferences

Can work on-site

Paris (up to 50km)

Echo Analytics
Tech Lead Machine Learning Engineer
January 2024 - January 2025 (1 year)
Paris, France
Context:Echo Analytics transforms geospatial data into actionable insights. Echo provides businesses across all sectors with a powerful 2D digital map and visual polygons, offering data-driven analysis and insights for strategic and precise decision-making. Tech Lead for the Shape product, responsible for creating object polygons from satellite imagery.
Leadership
Managed a team of three ML engineers for the development of machine learning solutions for satellite imagery, particularly in object detection and segmentation.
Collaborated with the Product team to review and prioritize roadmap items and define the technical vision to ensure alignment with business objectives.
Promoted best practices in code and machine learning, mentoring junior colleagues.
Technical Contributions
Designed and implemented a machine learning pipeline for satellite image analysis with Dataflow, including Cloud infrastructure management.
Designed and developed a machine learning model training pipeline (SAM, GroundingDINO) with Kubeflow on VertexAI, ensuring monitoring and observability of data and performance.
Developed a prototype using LLM and RAG to translate natural language into SQL queries, enabling the Delivery team to efficiently customize deliveries for clients.
Developed an AI assistant prototype that verifies product feasibility, allowing the Product team to respond quickly to client requests.
Python (Programming Language) Pytorch Google Cloud Platform (GCP) PostgreSQL Docker CICD Kubeflow Vertex AI
Trimble
Research Engineer
October 2021 - December 2023 (2 years and 2 months)
Paris, France
Context:Bilberry was acquired by Trimble in 2022, ranking among the top 10 French acquisitions of the year. Bilberry's Intelligent Spot Spraying system reduces herbicide use by over 80% while protecting the environment and lowering costs for farmers. Member of the AI team, responsible for developing weed detection algorithms.
Leadership
Managed a team of over 25 annotators, and developed annotation guidelines to ensure consistency and adherence to project standards in all annotations.
Organized internal learning sessions, mentored junior colleagues, and participated in technical profile recruitment.
Technical Contributions
Designed and developed machine learning models for weed identification, optimized for embedded systems, with accuracy rates (80-90%) for different use cases.
Designed and developed a machine learning-assisted annotation system, reducing data annotation costs by 70%, saving hundreds of thousands of euros.
Developed data ingestion and selection pipelines capable of processing up to 10,000 images per day.
Created dashboards to improve monitoring of model performance and operational metrics.
Python (Programming Language) Pytorch Caffe TensorFlow Grafana PostgreSQL Google Cloud Platform (GCP)
TotalEnergies R&D
Research Engineer
ENERGY AND UTILITIES
April 2021 - September 2021 (5 months)
Palaiseau, France
Context:TotalEnergies OneTech is TotalEnergies' technology and innovation hub, accelerating progress in energy solutions through research, engineering, and digital transformation. Intern with the subject Satellite Imagery & AI: Estimation of solar photovoltaic potential from open data.
Technical Contributions
Developed methods for identifying photovoltaic (PV) installation areas from open-source data, completing a full proof of concept and obtaining an intellectual property patent.
Implemented a data ingestion pipeline for creating training datasets from open-source data.
Developed a roof segmentation model with 90% accuracy.
Designed and developed a roof material classification model.
Designed and developed a roof obstacle detection model.
Python (Programming Language) TensorFlow