Safouane Chergui

• Architected RAG pipeline for document parsing and Q&A; over internal knowledge bases • Led development of LLM evaluation framework for business-critical tasks • Managing a team of 4 data scientists

• Deployed a RAG system reducing customer support costs by 20% • Enhanced phone assistant with fast search system and agents for appointment booking

• Built LLM-powered catalog system reducing onboarding from 20+ days to hours • Deployed small LLMs reducing third-party API costs by 10X • Implemented MLOps pipeline with MLflow for automated model retraining

• Designed recommendation algorithms driving 35% increase in page visits • Pioneered LLM review summarization processing 500K+ reviews monthly • Built collaborative filtering models serving 20M+ users

• Deployed ML models for segmentation, churn (AUC 0.92), and LTV modeling • Designed real-time drift detection monitoring 15+ production models • Built A/B testing framework reducing analysis time by 60%

- Facilitated discussions and provided guidance to students in the online community forums - Contributed to the development of course content and curriculum refinement based on student feedback and learning outcomes

- Anomaly detection using signal processing & ML algorithms (CNN, RNN) - Measured the performance of implemented ML models on real life power plant installations - Written a paper in collaboration with LAMCOS – INSA Lyon