Scale Experimentation in ML-driven Drug Discovery, MLOps from Research Perspective at Novo Nordisk – Carsten Stahult, Novo Nordisk & Kjell Carlsson, Domino Data Lab

Session Outline

Developing machine learning models to discover new candidate peptides is critical for developing new treatments for diabetes and other chronic diseases, but how do effectively test molecules and train models across a nearly infinite set of possibilities. In this fireside chat session, we discuss how Novo Nordisk uses an active learning approach to maximize the effectiveness of its molecular testing and ML model development in tandem, and the value of MLOps capabilities — like experiment tracking — in scaling experimentation and accelerating research.

Key Takeaways:

• How to leverage an active learning approach to drive more targeted, effective experimentation and ML model development
• The importance of experiment tracking and reproducibility in ML-driven research
• The benefits of MLOps capabilities in facilitating ML experimentation and deployment at scale