Cloud Holds the Key to Faster Development of AI and Machine Learning

Billions of dollars are spent annually by data scientists and the companies they work for, developing artificial intelligence and machine learning models. But organizing and implementing them can be an extremely expensive and manual process consuming literally man-years of highly experienced developers’ time.

Adding further complexity, software firms may require teams working constantly to keep the layer up to date with the latest technologies.

The lack of AI infrastructure has become a “last mile” problem for companies, introducing an element of drag that can stop many AI/ML investments from ever reaching production.
Cloud Services Providers Raw HTML Module“Every company with an AI strategy is going to need to build their own AI layer, which would take years and cost millions of dollars – or buy one,” says Diego Oppenheimer, co-founder and CEO of Algorithmia, a Seattle-based marketplace for AI/ML models that has grown to support over 55,000 developers and 4,000 algorithms, functions and models, making it the largest AI platform in the world.

The company includes researchers from key higher-education institutions including MIT, the University of Washington and Carnegie Mellon University.

“Our view is that companies would rather buy a platform that works today, is supported by some of the world’s leading tech companies and AI experts, and costs a fraction of what it would take to build their own solution,” says Oppenheimer. “We already have several large enterprise companies using the Algorithmia AI Layer and look forward to partnering with all the major cloud providers to deliver the AI layer to their customers.”

The Algorithmia AI Layer offers two versions: Serverless and Enterprise. With the Serverless AI layer, an engineer or data scientist can create an account and add AI/ML models in the language of her choice, hosting the models within Algorithmia’s cloud. The Enterprise layer uses the Algorithmia team to deploy software to private or public clouds.

The flexibility of this approach means data scientists can add models, and engineers around their organization can discover and use these models via an API call. Authorization and permissioning can be customized to work with any organizational structure, while devops can be fully automated.

Algorithmia is already working with US government agencies, helping them to deploy new capabilities in their AI layer. The platform delivers the security, scalability and discoverability that their data scientists need to focus on problem solving.