SigOpt is a startup from San Francisco which develops the platform for optimization of models of the software on the basis of artificial intelligence.
History
2020: Intel purchased SigOpt developer
At the end of October, 2020 Intel announced SigOpt purchase. The transaction should be closed in the fourth quarter 2020, but its conditions are not disclosed. The American semiconductor giant is going to use technologies of a startup in the hardware solutions to help developers to develop the AI tools.
SigOpt specializes in the software on the basis of AI and machine learning. The CEO of the company Scott Clark based a startup in 2014 together with Patrick Hayes who worked as the software engineer in BlackBerry, Foursquare, Bloomberg and ppFacebook]]. The platform and API SigOpt optimize hyper parameters - parameters which values are used for process control of training of the AI model, - for creation of high-performance models in production. The SigOpt platform supports 100 hyper parameters and also Bayesian and global algorithms of optimization.
Using several code lines software developers will be able to build in SigOpt workflows irrespective of type of the AI platform, cloud infrastructure or client library. Using library which can create up to several tens of thousands experiments at the same time, SigOpt helps to parametrize models and to start optimization cycles, to trace metadata of experiments and to provide the analysis on a web analytics dashboard.
SigOpt claims that it works with the "leading" algorithmic traders and the government organizations and also with academic institutions. The startup attracted in total $8.7 million a venture capital from such investors as Andreesen Horowitz, SV Angel, In-Q-Tel and Blumberg Capital. After the conclusion of the transaction the SigOpt command will join department of performance improvement of machine learning of Intel.[1]