Johan Gerfaux

• • Designed and deployed a production-grade internal agentic LLM with LangGraph and FastAPI, now actively used by 1,000+ employees.

• • Built use cases enabling users to retrieve data from internal databases and run analysis through a secure conversational interface.

• • Collaborated with cross-functional teams to deliver a robust AI service, driving productivity gains and adoption across 4 business units.

• • Collaborated in a cross-functional Task Force to address a high-priority production issue by analysing hundreds of thousands of data points. Applied clustering (DBSCAN) to separate OK vs Defective products, driving root cause discovery.

• • Developed CatBoost models on thousands of factory features to detect product defects with a 95% recall, ensuring defective parts are not missed.

• • Partnering with production engineers to integrate predictive insights into daily operations.

• • Co-developed a production-focused LLM for root cause analysis, adopted by 200+ engineers across the company. Built scalable data ingestion with Databricks + PySpark pipelines.

• • Led a Lean Six Sigma Yellow Belt project, delivering a Streamlit app that reduced time to statistical insights by 90% for 30 users, driving adoption of data-driven problem solving.

• • Designed and deployed data infrastructure on AWS (Redshift, Athena, S3) + DBT, enabling real-time and aggregated dashboards (yields, scrap costs, defect trends) for management decision-making.

• • Supervised and mentored a summer intern on driving data-driven practices in Quality Engineering.