DynamoFL operates a platform that allows developers to train AI models on personally identifiable information (PII) without compromising privacy, at scale. The company aims to make AI development more accessible in industries where data privacy is critical, such as healthcare and financial services. DynamoFL is a graduate of Y Combinator’s 2022 winter cohort and was originally founded in 2021 by two PhD graduates from the Massachusetts Institute of Technology (MIT).
At its core, DynamoFL’s platform uses federated learning, a method that aggregates several smaller AI models trained by hundreds or thousands of different users in their own environments to create the final full-fledged model, removing the need to share sensitive data. Federated learning is, however, considered difficult to deploy, with several drawbacks including costs associated with transferring AI models back and forth, inaccuracy of the final model due to statistical variation across the component models’ data, and lack of personalization.
DynamoFL has developed a proprietary system, “FedLTN,” which avoids these drawbacks by using various techniques to prune AI models to a smaller size while maintaining accuracy. The platform also allows users to fine-tune models to specific cohorts.
Funding and financials
DynamoFL raised USD 15.1 million in a September 2022 Series A funding round co-led by Canapi Ventures and Nexus Venture Partners. The proceeds were funneled into new product development, and to grow DynamoFL’s research team.
By using this site, you agree to allow SPEEDA Edge and our partners to use cookies for analytics and personalization. Visit our privacy policy for more information about our data collection practices.