Jean-Baptiste Pettit

Design, development of deep learning algorithms applied to video post-production. Tasks include scene detection, object bounding box detection, context detection, object tracking over time. These problems are solved using state-of-the-art deep neural network algorithms (Siamask, TransNet,...) in Tensorflow 2 or PyTorch

Design, prototyping, production deployment, and team support based on internal needs for visibility into the company's operation of a big data pipeline for retrieving and normalizing asynchronous events from all company services to expose real-time KPI dashboards for leadership.

- Development of a Machine Learning platform adapted to very high volume data (several million predictions per second, several hundred terabytes of training data per day in the data lake)

- Design and development of prediction models (regression, classification, clustering) used at the core of company services

- Implementation of the architecture (data engineering) to test, train, and monitor model performance over time

- Writing high-requirement code (as it is directly intended for production) in Scala integrated into the core of critical company services

- Regular production deployment of new prediction services with monitoring and analysis (A/B testing) of their impact on the company's business

- Continuous improvement of prediction models through offline "feature engineering", online A/B testing.

- Strong involvement in product development, HR processes for ever-increasing needs for Machine Learning engineers and data scientists.

--------------------------------------------------------

- Developing a Machine Learning prediction platform on a big data infrastructure.

- Designing prediction (regression, classification, clustering) models from scratch used at the heart of the company.

- Implementing the data engineering architecture to test, train and monitor the models over time using Spark.

- Producing production level code in scala to use the prediction models within the critical services.

- Improving prediction models via offline feature engineering, online ab-testing, production deployment

- Heavily invested in product development, HR process for ML engineers / data scientists