Customer Stories: Startup creates platform for deploying machine learning models, brings it to Azure customers fast

This post was created by the Microsoft Customer Stories team and was originally posted on customers.microsoft.com

Smart deployment of machine learning models

The serverless AI Layer solution from Algorithmia automates DevOps for machine learning. It is available as a hosted service or as a customizable enterprise solution.

Normally, it would be difficult for data scientists to put their algorithms into production applications because they often don’t have the necessary skills or the right technology available. We help these scientists focus on their research by moving trained machine-learning models off their machine and into production automatically. Our platform runs private and proprietary algorithms alike and has access to more than 5,000 pretrained models in the AI Marketplace.

 

Customer Stories: Startup creates platform for deploying machine learning models, brings it to Azure customers fast

 

A scalable platform for real-time services

Many of our customers rely on the Algorithmia platform to run their models for them. That is, Algorithmia powers features of the live running services that these enterprises provide to customers, in real time. This approach works especially well with CPU- or GPU-based resources because we can handle millions of software calls from almost any kind of device or app—all at scale.

In addition to hosting models on Algorithmia.com, we have an enterprise solution (Algorithmia Enterprise) that brings the Algorithmia platform to enterprises’ local networks and to their choice of cloud provider. That means the algorithms (and the machine intelligence that they are a part of) are available and searchable by all developers within the company.

Broad support for cloud providers

The Algorithmia platform is cloud-agnostic. Enterprises devise different IT strategies, including multi-cloud strategies, and we know that adopting Algorithmia provides a valuable resource that complements the companies’ different types of workloads. With 90 percent of Fortune 1000 companies using Microsoft cloud solutions, it made sense to expand our market presence to include Microsoft Azure deployments and on-premises deployments. We see Microsoft for Startups as a great way to help bring that message to market.

The value of the cloud versus on-premises

Another key point is that we do not encourage customers to run solutions on-premises. However, we do want to support customers as they transition to the cloud. For customers who have no choice but to operate some solutions on-premises, Microsoft Azure Stack offers an exciting alternative. It brings Azure functionality to our customers’ on-premises datacenters and gives organizations more control over their data. Plus, it offers enterprises a chance to experience the value of the cloud before they’re ready to commit to it. Sometimes we have to do a lot of convincing as we evangelize the cloud, and having the prestige and respect that Microsoft brings to the table makes that task easier.

Strong support and a strong relationship

We’ve received excellent support from Microsoft during our participation in Microsoft for Startups—much better than we’ve experienced with other cloud providers. Microsoft provides fast responses to help us resolve technical issues and useful advice on how to better work with Azure. A prime example is how we were able to accelerate time to market for our Azure-supported version of our solution. We anticipated three or four months of effort to get Algorithmia ready for Azure, but with help from Microsoft for Startups, it took only a month.

Our goal is to make all applications smarter. With programs like Microsoft for Startups, we’re better able to build that community for the benefit of data scientists, developers, providers, and our business. That really is a win-win for all of us.

“We anticipated three or four months of effort to get Algorithmia ready for Azure, but with help from Microsoft for Startups, it took only a month.”

- Ed Blankenship: Head of Product

- Algorithmia