Gilles Madi, Ph.D. Email & Phone Number

Paris, Île-De-France, France

- 🏥 Develop advanced AI models for health and insurance: Design, build, and deploy AI and generative models to tackle complex challenges such as health risk assessment, claims processing, fraud detection, and personalized insurance solutions.- 🔬 Research and innovation in AI for health and welfare: Stay informed about the latest advancements in AI and Gen AI for health and welfare insurance.- 🧪 Experiment with new algorithms and technologies: Improve model performance and introduce… Show more - 🏥 Develop advanced AI models for health and insurance: Design, build, and deploy AI and generative models to tackle complex challenges such as health risk assessment, claims processing, fraud detection, and personalized insurance solutions.- 🔬 Research and innovation in AI for health and welfare: Stay informed about the latest advancements in AI and Gen AI for health and welfare insurance.- 🧪 Experiment with new algorithms and technologies: Improve model performance and introduce innovative solutions that enhance health outcomes and welfare services.- 📊 Health data analysis and modeling: Perform sophisticated analyses on health and welfare insurance data to extract meaningful insights.- 🤝 Collaborate with multidisciplinary teams: Work closely with underwriters, actuaries, healthcare professionals, and other stakeholders to integrate AI solutions into products and services.- 🔍 Contribute to the development and improvement of our internal AI-based fraud detection tool.- 📢 Communicate insights to stakeholders: Present complex analytical results clearly and concisely to both technical and non-technical audiences, including management and international clients.- ✅ Ensure data quality and compliance: Ensure all AI models comply with ethical standards, data privacy laws (e.g., GDPR, HIPAA), and international regulatory requirements (EU AI Act).- 🎯 Strategic contribution to business objectives: Help define the AI roadmap aligned with long-term business goals in the international health and welfare insurance market.- 💡 Maximize value for clients, partners, and stakeholders through innovative AI solutions. Show less

Paris, Île-De-France, France

🌟 Development and deployment of an AI-based computer vision solution for theft detection, resulting in a significant reduction in false alert rates and improved efficiency in identifying suspicious behavior.🌟 Spearheaded the research and implementation of state-of-the-art deep learning models, including VideoMAE and Vision Transformer (ViT), to analyze video data for theft detection purposes.🌟 Collaborated closely with cross-functional teams to integrate the developed models… Show more 🌟 Development and deployment of an AI-based computer vision solution for theft detection, resulting in a significant reduction in false alert rates and improved efficiency in identifying suspicious behavior.🌟 Spearheaded the research and implementation of state-of-the-art deep learning models, including VideoMAE and Vision Transformer (ViT), to analyze video data for theft detection purposes.🌟 Collaborated closely with cross-functional teams to integrate the developed models seamlessly into operational workflows, ensuring smooth deployment and adoption across the organization.🌟 Utilized innovative modalities such as RGB and skeleton data to enhance the accuracy and robustness of the detection system, enabling more comprehensive analysis of human behavior.🌟 Leveraged cloud computing platforms like GCP to deploy and serve the AI models, optimizing performance and scalability while adhering to stringent security standards.🌟 Developed robust analytics and reporting mechanisms to track key metrics and evaluate the effectiveness of the deployed models, providing valuable insights for ongoing optimization and improvement efforts. Show less

Basel, Switzerland

✓ Intelligent Document Processing.✓ Designed, developed and deployed a separation Head for intelligentdocument separation, resulting in a full automatisation of the processfor the a bank company. -Language models Lilt -Topic Modeling, Latent Dirichlet Allocation✓ Designed, improved and deployed graph neural network models forintelligent document processing serving top tier manufacturing andbanking companies.✓ Deployment and serving through GCP.✓… Show more ✓ Intelligent Document Processing.✓ Designed, developed and deployed a separation Head for intelligentdocument separation, resulting in a full automatisation of the processfor the a bank company. -Language models Lilt -Topic Modeling, Latent Dirichlet Allocation✓ Designed, improved and deployed graph neural network models forintelligent document processing serving top tier manufacturing andbanking companies.✓ Deployment and serving through GCP.✓ HuggingFace, PyTorch, Docker, Cloud functions, Flask, REST, Git, clickup, analytics, reporting, tableau, Show less

Paris Area, France

✓ Performing research and development at the forefront in machinelearning and deep learning to find algorithms, improvements andoptimizations of the performance of predictive models on issuesrelated to time series, vision, regression, classification and to the automl.✓ Development, deployment and maintenance of prevision.io Auto-MLplatform serving +100k monthly users✓ Designed and developed a meta learning model for accelerating hyperparameter tuning of prevision.io… Show more ✓ Performing research and development at the forefront in machinelearning and deep learning to find algorithms, improvements andoptimizations of the performance of predictive models on issuesrelated to time series, vision, regression, classification and to the automl.✓ Development, deployment and maintenance of prevision.io Auto-MLplatform serving +100k monthly users✓ Designed and developed a meta learning model for accelerating hyperparameter tuning of prevision.io auto ML platform -Bayesian search, TPE, PCA -Constraint Programming, Java, Choco Solver -Meta-learning, Few-shot learning, cold/hot start training✓ Designed a time series forecasting model for an energy company thatimproved by 25% their daily energy consumption forecast. -Attention mechanism, Heat-map, Auto-lag values -Keras, Transformers model, LSTM Show less

Creation and implementation of an analysis and learning system for prediction of deviant behavior in virtualized data centers.- Modeled historical workloads using global time seriesconstraints and associated to a- predictive LSTM model. ____Création et mise en place d'un système d'analyse et d'apprentissage pour la prédiction de comportements déviant dans les centres de données virtualisés.

Nantes Area, France

Combining constraint programming and machine learning to come up with an energy aware model for small/medium size data centers.Over the last decade, cloud computing technologies have considerably grown, this translates into a surge in data center power consumption. The magnitude of the problem has motivated numerous research studies around static or dynamic solutions to reduce the overall energy consumption of a data center. The aim of this thesis is to integrate renewable energy… Show more Combining constraint programming and machine learning to come up with an energy aware model for small/medium size data centers.Over the last decade, cloud computing technologies have considerably grown, this translates into a surge in data center power consumption. The magnitude of the problem has motivated numerous research studies around static or dynamic solutions to reduce the overall energy consumption of a data center. The aim of this thesis is to integrate renewable energy sources into dynamic energy optimization models in a data center. For this we use constraint programming as well as machine learning techniques. First, we propose a global constraint for tasks intersection that takes into account a ressource with variable cost. Second, we propose two learning models for the prediction of the workload of a data center and for the generation of such curves. Finally, we formalize the green energy aware scheduling problem (GEASP) and propose a global model based on constraint programming as well as a search heuristic to solve it efficiently. The proposed model integrates the various aspects inherent to the dynamic planning problem in a data center : heterogeneous physical machines, various application types (i.e., ractive applications and batch applications), actions and energetic costs of turning ON/OFF physical machine, interrupting/resuming batch applications, CPU and RAM ressource consumption of applications, migration of tasks and energy costs related to the migrations, prediction of green energy availability, variable energy consumption of physical machines. Combining constraint programming and machinelearning to come up with an energy aware model forsmall/medium size data centres.✓ Predictive analysis and Statistical modeling✓ Data visualisation and reporting✓ ML Algorithms✓ Python, Java, Prolog, SQL, C++, Choco Solver,Sicstus soler, LaTEX Show less

Montpellier Area, France

Develop and implement traversing algorithms for multi-valued decision trees, while taking into consideration aggregation constraints over different valuation type. Problem of product configuration, with application in agriculture.

Montpellier Area, France

Develop and implement traversing algorithms for multi-valued decision trees, while taking into consideration aggregation constraints over different valuation type. Problem of product configuration, with application in agriculture.

Montpellier Area, France

Improve the decision making of an artificial intelligence for the GO, using a combination of the UCT / RAVE algorithms to choose the tree branch to explore.

Doctor Of Philosophy (Ph.D.) - École Des Mines De Nantes, Computer Science

Activities and Societies: Combining constraint programming and machine learning to come up with an energy aware model for small/medium.

Master'S Degree - Université De Montpellier 2, Computer Science

Bachelor Of Science (B.Sc.), Maths And Computer Science