Brice De La Brière Email & Phone Number

Paris, Île-De-France, France

Paris, Île-De-France, France

• Led and developed a team of machine learning engineers, overseeing recruitment and onboarding while cultivating a culture of continuous learning and career advancement to enhance the team's expertise and innovation in machine learning.• Provided technical leadership on all ML and AI projects, with a strong focus on user experience personalization and recommender systems, achieving outstanding results.• Responsible for the maintenance and performance optimization of ML models, ensuring their long-term reliability and effectiveness. Implemented an on-call system within the team to ensure timely responses to model performance issues.

Paris, Ile-De-France, France

• Improved the video categorization pipeline with a new SOTA model for features extraction (EfficientNetv2) and a multimodal approach using audio signal and metadata. • Developed a video categorization pipeline using deep learning techniques used at scale in production. Wrote a medium post about the project for Dailymotion's blog: ”How Deep Learning can boost Contextual Advertising Capabilities”. (InceptionV3, NeXtVLAD, TensorFlow, Dataflow, Docker, Kubernetes, TensorFlow Serving).

Paris Area, France

• Developed a causal inference tool to better analyze Dailymotion's data. The tool Implements famous matching methods (Mahalanobis Distance Matching (MDM) and Propensity Score Matching (PSM)), Coarsened Exact Matching (CEM), and Instrumental Variables: Two-Stage Least Squares (2SLS).• Various contributions and improvements on the main dailymotion's video recommender. (Recurrent neural network RNN, counterfactual reasoning, Tensorflow, Tensorflow serving)• Realized a recommender system for dailymotion’s website, the system dynamically ranks sections given some context information, optimizing for our KPI. Based on deep contextual bandits methods (Airflow, Beam, TensorFlow, TensorFlow Serving, Docker, etc.). Offline policy evaluation via propensity scoring.

Paris, Île-De-France, France

• Developed a fraud detection solution for ad fraud over dailymotion’s video player (Airflow, Pandas, scikit-learn). Model’s performance monitoring (Datadog and Tableau). Developed an API (Flask, Docker, Kubernetes) to serve the model at scale.

Région De Paris, France

London, Royaume-Uni

Master'S Degree, Artificial Intelligence, Systems, Data (Iasd)

Engineer'S Degree, Computer Science

Cpge (Psi), Classes Préparatoires Aux Grandes Écoles