SigOpt: Elevating Business Through Optimization
Scott Clark, a Cornell graduate with a PhD in Applied Mathematics and a Masters in Computer Science, views optimization as the key to success in organizations. In fact, he sees many organizational efforts as an optimization challenge. Organizations spend a lot of time attempting to discern which set of inputs will provide the most valuable set of outputs. For example, a cookie company might be unsure as to what amount of vanilla is needed for their top-selling item. To solve this problem, Clark started SigOpt in 2014 to offer optimization as a service. The company collects data to predict what would occur to an end result if a variable is altered in the process. Dr. Clark shares, “Our black-box optimization engine takes in information about configuration parameters and how they performed but all information about the internal model, data, and company are abstracted away”.
Since its launch, SigOpt has made great strides in the technology industry. The company prides itself on its ability to optimize in a variety of ways. SigOpt has successfully optimized advertisements to chemical formulas for products. In fact, the consumer packaged good and financial services industries have been most inclined to implement SigOpt in their practices; however, Clark hopes to eventually offer a self-serve product. Dr. Clark says, “In general, we’re 10-100 times faster than normal techniques (this saves time and computational cost) and produce results that are significantly better. Due to its innovative offerings, Clark has secured funding with a series of investors such as Y Combinator and Data Collective. In 2016, Clark was named to Forbes’ ‘30 under 30’ list in Enterprise Technology.
Scott Clark shared more about his business with Entrepreneurship@Dyson below:
What is SigOpt doing differently from its competitors?
SigOpt is an optimization platform that seamlessly tunes Artificial Intelligence (AI) and Machine Learning (ML) model parameters through a state-of-the-art ensemble of Bayesian and global optimization algorithms behind a simple Application Programming Interface (API). This results in captured performance left on the table by conventional techniques and reduces the time and cost for developing and optimizing models. We differ from competitors because we can seamlessly bolt on top of existing pipelines and provide an additive boost by tuning competitors’ configuration parameters and never needing to see the underlying data or model itself. This allows us to quickly boost performance to existing systems almost immediately.
What has been instrumental in the company’s success thus far?
Building a world class team of humble experts. We are constantly learning from each other and inspiring each other to build better and better solutions for our customers.
What has been the greatest challenge?
Finding the markets like financial services and oil and gas where our solution can provide immediate, tangible business value by optimizing the modeling processes that are tied to business value in these industries.
What is next for SigOpt?
Continuing to build our team and provide the best possible optimization to our customers around the world.
To learn more about SigOpt, please visit www.sigopt.com