Case Study: Exscientia reduces model monitoring and debugging time from days to hours with Seldon Deploy

How Exscientia reduced the time it takes to monitor and prepare models from days to hours

Exscientia, an AI-powered drug discovery company, needed to manage an automated pipeline that creates, deploys, monitors and retrains thousands of models without human intervention. As their scale grew, the existing open-source deployment and inference tools could no longer deliver the enterprise-grade observability, debugging speed and stability required, so they evaluated and selected Seldon and its Seldon Deploy product.

Seldon implemented Seldon Deploy to provide modular observability and production management that required no changes to Exscientia’s model code, allowing the team to port all models quickly. The Seldon solution cut time-to-debug for trivial issues from hours to minutes and reduced serious-issue resolution from roughly 24 hours to about one hour on average, speeding decision-making across data science and drug chemistry and accelerating oncology drug development.

Exscientia

Sash Stasyk

MLOps Engineering Team Lead