Louis-Melchior Giraud
Data Scientist & AI Engineer
I build reliable AI, data and software solutions for real-world use cases.
A practical mindset for AI & data that ships.
I am Louis-Melchior Giraud, a young Data Scientist and AI Engineer with experience in Earth Observation, machine learning, computer vision, dashboarding and full-stack development. I combine technical skills, clear communication and a practical mindset to build tools that are useful, understandable and production-oriented.
More about me, in motion.
In 90 seconds, discover who I am, what I build, and how I can help you with AI, data and software projects.
Organizations I have worked with
What I can help you with.
Eight focused areas where I deliver reliable, production-oriented work.
Artificial Intelligence
Designing AI systems that are robust, useful and aligned with real business outcomes.
Machine Learning
Classical ML and deep learning models — from prototyping to deployment-ready code.
Computer Vision
Image segmentation, classification and detection on satellite and operational imagery.
Earth Observation
Geospatial pipelines for land-cover, cloud and high-resolution remote sensing tasks.
Data Engineering
Reliable pipelines: SharePoint, SQL, geospatial formats and reproducible workflows.
Dashboards & DataViz
Qlik dashboards and custom visualizations that turn raw data into clear decisions.
Full-stack Development
Modern web apps with React, Angular, Node.js, FastAPI and clean APIs.
Video & Communication
Technical video content and demos to make complex projects understandable.
Selected work.
A mix of professional, research and personal projects across AI, Earth Observation, dashboards and full-stack development.
Re-Identification of Graphic Assets
Deep learning model combined with a knowledge graph to re-identify graphic assets across visual collections. Bridges computer vision feature extraction with structured knowledge for robust matching.
Drug Consumption Prediction
Machine learning pipeline to predict drug consumption from socio-demographic and behavioral features, with model comparison, evaluation and interpretability.
High-resolution land-cover segmentation
Prototyped semantic segmentation models on high-resolution satellite imagery to produce detailed land-cover maps for Earth Observation use cases.
Cloud segmentation benchmarking
Benchmarked multiple deep learning architectures for cloud segmentation on multispectral imagery, measuring quality and inference cost.
TerraMind foundation model adaptation
Adapted the TerraMind foundation model to specific Earth Observation downstream tasks, with a focus on transfer learning and evaluation.
Explainable AI visualizations
Built explainability and visualization tools to help stakeholders understand model predictions on geospatial and tabular data.
Qlik dashboard for operational monitoring
Designed and built a Qlik Sense dashboard from SharePoint data sources to monitor operational KPIs for the Marine Nationale.
Attrition prediction with machine learning
Built ML models (SVM, CatBoost) to predict early resignation, with SHAP-based explanations and a clean, decision-oriented reporting layer.
Anki-like language learning app
Personal full-stack project: a spaced-repetition language learning app inspired by Anki with a modern UX and clean architecture.
Clothes renting full-stack app
End-to-end full-stack web application for clothes renting: authentication, catalog, booking flow and admin dashboard.
Professional timeline.
Working on production-oriented data science projects: real-world machine learning applications, data-driven decision support, robust and explainable models. Some projects are confidential and involve concrete business use cases and operational constraints.
High-resolution land-cover segmentation (U-Net + EfficientNet-B4), cloud segmentation benchmarking (OmniCloudMask, Dynamic World), TerraMind foundation model adaptation for multi-modal Earth Observation, explainable AI and uncertainty visualization tools. Collaborated with the GeoPulse team on rapid prototyping and deployment.
Built a Qlik dashboard with SharePoint API integration, data transformation and iterative co-design with end-users. Developed attrition prediction models (SVM, CatBoost, PyTorch) with feature engineering, SHAP-based interpretability and decision-oriented visualizations.
What people say about my work.
Honest feedback from people I have worked with, helped, or built something for.
“Louis-Melchior demonstrated technical skills typically seen in employees with several years of experience, as well as a business acumen that many never fully develop in their careers. He far exceeded expectations, showcasing both remarkable human and technical abilities.”
Tools of the trade.
A curated stack covering AI/ML, geospatial, data engineering, web and software development.
AI / ML
GEOSPATIAL
DATA & DB
WEB & BACKEND
SOFTWARE ENG.
TOOLING
Talk to my AI assistant.
A professional assistant that answers your questions about Louis-Melchior's profile, experience, projects and the kind of missions he can take on.
POWERED BY GPT-5
Let's build something reliable.
Whether you have a clear scope or just an idea, I am happy to discuss whether and how I can help. Expect a clear, professional answer.