Deci AI

Deci AI provides an end-to-end deep learning development platform that enables AI developers to build, optimize, and deploy more efficient and accurate models for various environments, including cloud, edge, and mobile. Founded in 2019 by Dr. Yonatan Geifman, Professor Ran El-Yaniv, and Jonathan Elial, the company was established to address the AI efficiency gap, where hardware struggles to meet the computational demands of increasingly complex models. Geifman serves as CEO, Elial as COO, and El-Yaniv as Chief Scientist.

The company's core technology is its proprietary Automated Neural Architecture Construction (AutoNAC) engine. This algorithmic optimization engine utilizes Neural Architecture Search (NAS) to redesign a trained model's architecture, enhancing its inference performance in terms of throughput and latency while preserving accuracy for specific target hardware. The platform is designed to shorten development cycles and reduce computing costs by up to 80%. Deci's offerings include highly efficient foundation models for computer vision, NLP, and Generative AI, as well as tools for customizing, optimizing, and deploying models into production. The company has released several models, including DeciLM-7B, DeciCoder, and Deci-Nano, which are designed to be smaller and more cost-effective than comparable models.

Headquartered in Ramat Gan, Israel, Deci AI serves clients across various sectors, including automotive, manufacturing, retail, cybersecurity, healthcare, and finance. Prior to its acquisition, the company had raised a total of $55.1 million in funding over three rounds, with its Series B in July 2022 raising $25 million led by Insight Partners. In May 2024, Nvidia acquired Deci AI for an estimated $300 million. This acquisition aims to complement Nvidia's AI expertise with Deci's capabilities in automated model design and performance improvement.

Keywords: deep learning platform, model optimization, neural architecture search, AI development, inference performance, AutoNAC, generative AI, computer vision, large language models, model deployment, edge AI, AI efficiency, cloud AI, model compression, AI acceleration, Deci-Nano, DeciLM, DeciCoder, MLops, AI algorithms