Striveworks Raises Initial Funding of $33M to Build Machine Learning Operations Tools

Image credit: Lagarto Films/Getty Images

Interest in MLOps (DevOps for people working with machine learning models) has boomed in the last year, and it’s not surprising. Organizations want to bring more machine learning into their data science stack, but it takes building. Train the model, clean the data, and make sure the model works. Today, a startup called Striveworks, which builds MLOps tools to handle that work, announced $33 million in funding.

This is the company’s first external funding, and the closing of this round highlights both the growing interest in the broader area of ​​artificial intelligence, as well as Striveworks’ own traction within it. The company’s ARR has increased by 300% year-on-year compared to the past year. 2 years.

The $33 million comes from a single investor, Centana Growth Partners, and Austin, Texas-based Striveworks will use the money for hiring and further product and business development. This funding was made in what is often described as an opportunistic round. Striveworks has been in business for his five years and has operated as a “capital efficient” start-up that has made a profit and invested it in growth, says company CEO Jim Levesco. – Founded the company with Craig Desjardins, Eric Corman and Tony Manganiello.

Levesco declined to name its current customers, but they include government and financial sectors that use machine learning to build services and run businesses, as well as “highly regulated industries, national security applications, and related fields such as computer vision orientation,” he said. , satellite imagery, commercial imagery, etc.,” he added. The company also partners with AWS and Azure to process data in those clouds. (Notably, it currently does not have a similar partnership with Google.)

The problem the company is tackling is one that Levesco, a neuroscience PhD and former longtime employee at financial services firm Virtue, and his co-founders regularly encountered at their previous company. and that Striveworks aims to combat this problem by being pragmatic in nature.

He said it starts with what he described as a “day-one problem”: how to build the right machine learning model for your goals. But that’s kind of the easy part. After that, the real intricacies begin.

“Does it work as expected, and will it continue to perform as expected when you put it into production?” he said. “We are focused on what happens next. “

Levesco describes himself as a “failed physicist” (I think it’s a reference to his pre-doctoral work) and learned an important lesson about AI models. All AI models are statistical and therefore fail. “So one of the key elements of responsibility is not only knowing that errors will occur, but putting in place automated and thoughtful plans to deal with them.”

He believes that as the use of AI becomes more prevalent, this will increasingly need to be considered. “Data models, AI models, ML models are becoming more and more important, not ad hoc models. Whether it’s credit scoring or healthcare, these databases are stored and queried. But how do you query [effectively] Are you so wrong? “

The company aims to tackle this problem through its flagship platform called Chariot. Chariot can be used to prepare data, build models, and run those models in production. Features on the platform that use a low-code format suitable for team collaboration include model-in-the-loop annotations, the ability to import models and use previously cataloged data models (from your own organization), custom It includes the ability to build It features workflows, queries for the “source” of data in sets, integration with third-party tools, and more.

There are currently many start-ups (and large companies) on the market working on MLOps solutions. Examples we have covered include Seldon, Galileo, Aries, and Tecton. Major system integrators are also joining the movement, such as McKinsey’s recent acquisition of Yguagio.

Ben Cukier, a Centana partner who led the investment, said that Striveworks has a clear advantage over these in that the business itself is running very well, and that this is a big part of how the company operates and what they achieve. It is a sign of both.

“They have reached triple digit growth when they were just a Series D. I was amazed to see how efficiently they used their capital. And I’ve only seen a few companies that can achieve this scale without external capital, which is a rare occurrence.These are genuine customers with seven-figure contracts, and many other We have net retention numbers that companies would envy.”

The company didn’t disclose a valuation, but Cuquier described the current market as not “calm” but simply “normal,” a return to business as usual after a very tumultuous few years.