Algorithmia Aims AI Layer at ‘Last Mile’ Problem
It has been asserted that the algorithm has overtaken traditional software code as the driving force behind the AI Era. On the assumption that assertion is accurate, startups are beginning to offer tools for building what one refers to as “algorithm portfolios” used to deploy AI and machine learning models in enterprise applications.
Among them is application builder Algorithmia, which this week released an “AI Layer” designed to accelerate the development and deployment of machine learning models. The Seattle-based startup claims its AI Layer fills a gap in the current AI infrastructure that prevents AI and machine learning investments from reaching production.
Algorithmia notes that developing and implementing machine learning models has become an arduous and expensive task. Model developers still face a “last mile” to deployment problem for lack of AI infrastructure, it notes.
The company’s API is touted as enabling data scientists to automate deployment of machine learning models to production using TensorFlow or other open source machine learning libraries.
“Tensorflow is open source, but scaling it is not,” Kenny Daniel, co-founder and CTO of Algorithmia, asserted in releasing the tool on Thursday (Nov. 16). “Almost all R&D has focused on collecting and cleaning data, and building models. Algorithmia has spent the last five years building the infrastructure that will put those models to work.”
The AI Layer is aimed primarily at enterprise customers, but also comes in what the startup calls a “serverless version. In the latter, data scientists can use Algorithmia’s API to develop models hosted in the startup’s cloud.
The enterprise version of the AI Layer allows models to be deployed in any public or private cloud, hence, the company notes, creating the ability to scale machine-learning models. Once software in deployed in the cloud, developers can call on the API to access models or add new ones.
The startup’s goal of bridging the last mile between algorithm development and production has attracted some impressive backers, including Google’s (NASDAQ: GOOGL) vice president of AI engineering and In-Q-Tel, the CIA’s investment arm. “As someone [who] has spent years designing and deploying machine learning systems, I’m impressed by Algorithmia’s serverless micro-service architecture,” Google’s Anna Patterson stated in a testimonial.
In its efforts to push machine-learning algorithms out of the lab and into production, Algorithmia claims its portfolio of machine learning models is the largest AI marketplace of its kind. The platform on which its new AI Layer is based includes algorithms and models from Caltech, Carnegie Mellon University, the University of California at Berkeley, MIT and others.
Late last year the startup released a machine image running on the Amazon Web Services (NASDAQ: AMZN) cloud along with a pipeline for creating style transfer models. The tool allows users to set up a deep learning environment in the cloud, train models and release them as a scalable REST-based API. The platform allows users to train models in about a day using GPU-based Amazon Elastics Compute Cloud instances.