Case Study: Recursion Pharmaceuticals accelerates drug discovery with Looker

AI-enabled pipeline finds treatments for rare diseases

Recursion Pharmaceuticals, a life sciences company using AI and experimental biology to speed drug discovery, needed a way to handle rapidly growing volumes of microscopy images and machine learning workloads. As its high-throughput labs scaled, it was producing compute- and memory-intensive data at a pace that strained its existing batch-processing approach, so it looked to Looker-supported Google Cloud tools and services, including Kubernetes-based infrastructure and analytics workflows.

To address this, Recursion Pharmaceuticals implemented a hybrid cloud platform with Google Kubernetes Engine, Confluent Kafka, BigQuery, Cloud Storage, and Cloud TPU pods for training and inference, helping streamline image processing and model development. The results included reducing deep learning training time from over 24 hours to as little as 15 minutes, accelerating microscopy image processing, and enabling hundreds of rare-disease models and drug candidates, including one of the first machine-learning-discovered compounds to reach Phase 1 clinical development.

Recursion Pharmaceuticals

Ben Mabey

Vice President Engineering