Arize AI, a startup developing a platform for machine learning operations, today announced it has raised $38 million in a Series B round led by TCV in partnership with Battery Ventures and Foundation Capital. Raising Arise’s total capital to $62 million, CEO Jason Lopetaki said the new cash will be used to scale R&D and double the company’s 50-person headcount over the next year.
Machine learning operations, or MLOps, involve implementing and maintaining machine learning models in production. Similar to DevOps, MLOps aims to increase automation while improving the quality of production models – but not at the expense of business and regulatory requirements. Given the interest in machine learning and AI in the enterprise, it’s no surprise that MLOps is expected to become a large market, with IDC forecasting the size to reach $700 million by 2025.
Arise was founded in 2019 by Lopetkey and Aparna Dinakaran, after Lopetkey’s previous startup – Tubemogul – was sold to Adobe for around $550 million. Lopeteki and Dhinakaran first met at TubeMogul, where Dinakaran was a data scientist before joining Uber to work on machine learning infrastructure.
“After a year of seeing team after team — we came to the conclusion that something was fundamentally missing — the models that were delivered to production failed to understand what was wrong, and once the models failed to understand what the models were doing,” Lopatecki said at Technology Flow. An email interview. “If the future is AI-driven, there needs to be software that helps humans understand AI, break down problems and solve them. AI without machine learning observability is not sustainable.
Arise is certainly not the first to tackle these kinds of challenges in data science. Another MLOps vendor, Tecton, recently raised $100 million to build its machine learning model experimentation platform. Other players in the space include Galileo, Modular, Gantry and Grid.i, the latter of which raised $40 million in June to launch a component gallery that adds AI capabilities to apps.
But Arise is unique in many respects, Lopatecki says. The first is a focus on observability: Arise’s product of embeddings is designed to look inside deep learning models and understand their structure. “Bias tracing” complements that, a tool that monitors biases in models (eg, facial recognition models that detect black subjects less often than light-skinned subjects) – and tries to rediscover the data that causes the bias.
Recently, Arise launched Embedding Drift Monitoring, which tries to detect when models become less accurate as a result of outdated training data. For example, “Who is the current US president?” Drift monitoring raises alerts the customer if a language model answers “Donald Trump” in response to the query.
“Arise stands out… [because] We’re laser-focused on doing one hard thing well: machine learning observability,” says Lopetecki. “Ultimately, machine learning infrastructure looks like a software infrastructure with many market-leading, best-of-breed solutions that machine learning engineers use to build great machine learning. We believe.”
Arize’s second differentiator, Lopatecki says, is its domain expertise. He and Dhinakaran both come from academia and draw from practitioner roots, he noted — building machine learning infrastructure and managing problems with models in production.
“Even for teams of experts and thought leaders, it’s becoming impossible to keep up with every new model architecture and every new advancement,” says Lopetecki. “As soon as teams finish building their latest model, they’re usually jumping on the next model the business needs. This leaves little time for deep introspection about the billions of decisions these models make every day and the impact these models have on businesses and people…hence the need for deep learning models and a tool to monitor designed workflows. Arise spent over a year building the product. To troubleshoot where they go wrong.”
Some might (correctly) argue that Arise’s competitors have experts in their ranks and surveillance and monitoring solutions in their product suites. But judging by Arise’s impressive client list, the startup is making a convincing sales pitch. Uber, Spotify, eBay, Etsy, Instacart, P&G, TransUnion, Nextdoor, Stitch Fix and Chick-fil-A are among Arise’s paying customers, and the company’s free tier — which launched earlier this year — has 1,000 users.
However, mum’s word on the annual recurring revenue. Lopatecki insists that the capital from the Series B will give the company “ample runway,” improving the macro environment.
“In healthcare, there are teams using Arize to ensure that cancer detection models using images are consistent in production across a wide range of cancer types. In addition, there are teams using Arize to ensure models used in standards of care decisions and insurance experience are consistent across ethnic groups,” added Lopetecki. “As models become more complex, we’re seeing even the largest and most sophisticated machine learning teams realize that they need to invest their time and energy in building better models rather than building a machine learning observability tool … Arise helps learners improve return on investment in models and quantify results for business leaders. [and provides] Market-leading software for monitoring the risks of AI investments.