About the Client
Huma is a global healthtech company transforming healthcare through digital biomarkers and therapeutics. Huma has raised a total of $54M in funding and acquired 3 organizations since inception.
The Challenge
Cloud XLR8R were tasked with delivering a scalable machine learning platform to facilitate repeatable workflows and fast experimentation. Huma had an existing application they wanted to migrate to the platform that could be used as a reference for onboarding other projects.
The Solution
Cloud XLR8R deployed Kubeflow, a scalable, cloud agnostic ML platform built on top of Kubernetes. Cloud XLR8R assisted in migrating Huma’s existing workflows to the platform and added triggers to run them when new data arrived or the codebase was updated. In addition, CI processes were set up for building, testing and deploying these. Workshops and drop in clinics were held to train the team on using the platform and to take advantage of advanced features such as GPU autoscaling.
Business Outcomes
Huma gained a repeatable process for running their workflows which enabled greater collaboration between team members. Having rigorous CI and CT pipelines gave them greater confidence in deploying to production, reducing their time to market. Having a modern ML platform with features such as autoscaling saved costs and sped up development of new models.
Customer Testimonial
“Cloud XLR8R worked with us to understand our needs in detail and then help construct a robust cloud machine learning infrastructure that we could deploy across multiple projects, using modern container and pipeline platforms, fully automated from a devops perspective. Most importantly, they collaborated really closely with the team, reacted to changing requirements, ensured great handover and training and were always very professional. We expect to use and build on the infrastructure for some big initiatives over the next year, and anticipate it will speed development, ensure reproducibility and reliability and allow us to scale up and down as needs change.”