SigOpt Inc. works in a data analytics discipline that is so specialized that few people outside of the data science field can even understand it, much less explain it. That hasn’t stopped the eight-person venture from raising a $6.6 million series A funding round led by blue-chip Silicon Valley venture capitalist Andreessen Horowitz for its “optimize everything” cloud platform. That brings total funding for the two-year-old company to $8.8 million.

SigOpt’s specialty is Bayesian optimization, a “sequential design strategy for global optimization of black-box functions that doesn’t require derivatives,” according to Wikipedia. The technique is commonly used in machine learning applications to reduce the number of parameters the algorithm needs to consider. The company says its cloud service is being used at everything from hedge funds to breweries to shortcut the trial-and-error process, with typical time savings of more than 90 percent. An explanation of Bayesian optimization that is accessible to most humans is here.

CEO and Co-founder Scott Clark (@DrScottClark, above) came upon the idea while pursuing his Ph.D. studies at Cornell University, then tested it in ad targeting at Yelp Inc. The technique shortcuts the labor-intensive testing process by examining test data and recommending which experiments to run next, which reduces trial-and-error on the part of data scientists.

“Traditionally, when data scientists are building a machine learning pipeline, there are a lot of tunable knobs and levers,” Clark said. “SigOpt helps them find the best filters to use. It guides them to the best combination.”

Given that data scientist salaries start in the six figures these days, SigOpt’s pricing, which starts at $1,000 per month for 10 experiments and goes down from there, is a relative bargain. “In some industries, half the time of a data scientist is spent tuning and tweaking,” Clark said. “We often see cost reductions in computation resources of 90 to 95 percent.”

Customers include MillerCoors LLC, Prudential plc and Johnson & Johnson, as well as “some of the largest banks, hedge funds and insurance companies in the world,” Clark said.

The company’s cloud service uses a REST (representational state transfer) application programming interface (API) that developers can plug into to access its algorithms. Customers then feed meta-information and parameters into the system and specify a goal. The product is integrated with Google’s TensorFlow, SAP HANA, R, Java, Python and scikit-learn machine-learning programs.

Although hiring any data scientists – much less those specialized in Bayesian optimization – would seem to be a challenge these days, Clark said it hasn’t been a problem for his company. “We’ve been fortunate to have a healthy pipeline of people,” he said. SigOpt plans to use the funding to roughly double its staff and invest in engineering and sales. Other investors include Data Collective, SV Angel, Stanford University and Blumberg Capital.

Image courtesy SigOpt