Scaling a Machine Learning business

Posted By:

I run a data science / advanced analytics practice. Our opportunities are abundant, the thrills are a dozen-a-minute, and the joy of driving to a successful business outcome for my clients makes it all worthwhile.

However, this business is not for the faint of heart. It is fast-paced and dependent on highly skilled resources … this makes scaling the practice challenging. We’ve all heard about the exploding need for advanced analytics professionals and the concomitant scarcity of credentialed data scientists.

When Neal Analytics was offered the opportunity to participate as an early adopter of Passau (now called Microsoft Azure ML), we were intrigued. While Microsoft knows how to build great point-and-click/drag-and-drop products, I wasn’t sure what this meant for the art of data science. After 9 months of kicking the tires with Azure ML, I believe this product has potential.

Collaboration: Azure ML allows for effective teaming through shared workspaces and a light workflow. The data science activities can be divided into blocks that experienced and junior data scientists can divide and conquer.

Speedy time-to-insight: With visual composition and out-of-box models, our analytics professionals can be up and running in a matter of hours rather than the previous days or weeks.

Cloud-scalability: Data-intensive projects previously imposed an additional burden around the management of data compute infrastructure. With Azure ML, we can scale up to large compute instances without the lag from the setup of big data infrastructure and at a better TCO.

Azure ML is still a V1 product and has room to improve. But features like the above combine to deliver an effective force-multiple for my highly-skilled data science talent. We hope to go higher, further, and faster with Azure ML.

I’ll be speaking about our experiences with the product at WPC 2014 on 7/15. I look forward to seeing some of you. Cheers!