Thabet Chaaouri

Leading the delivery and deployment of ML solutions at scale in FTV’s production environments : • Led a team of 2 data engineers and 1 ML engineer to deliver production-grade AI solutions. • Scaled an open-source Speech-to-Text solution across France Télévisions, enabling automatic transcription for newsrooms and archives. • Delivered transcription and translation workflows that supported FTV’s exclusive coverage of the Paris 2024 Olympics. • Designed and deployed end-to-end support chatbots and autonomous agents using LLMs, LangGraph, and Langfuse (for monitoring), deployed via Kubernetes in secure environments.

Developing intelligent tools tailored to the needs ofofnewsrooms : • Collaborated with journalists to understandunderstandnewsroom challenges and designed ML solutions accordingly. • Developed end to end ML demos to showcase solutions to business stakeholders • Designed a topic modeling pipeline ( using Bertopic ) on FTV news stories to promote collaboration between journalists across different newsrooms. • Integrated AI solutions to: - support fact checkers in “Fake News” detection by implementing an image similarity system and deep learning models to detect Deep Fakes. - Improve archiving workflows with STT models. - Automate music authorship declaration via a music recognition system • Experimented with fine tuning opensource LLMs for various use cases (News recommendation , Support,…)