Why we invested in DynamoFL, a federated learning platform for AI

The need for privacy-protected artificial intelligence (AI) applications is growing as machine learning becomes more commonplace across industry verticals.

DynamoFL has developed a solution that enables developers to train highly personalized machine learning models without ever having to collect user data, and without having to trade-off high performance. It uses sophisticated optimization methods, in which each client gets a model that is good for the overall task – a task that all the other clients are interested in – and that also does particularly well in the personal domain the client exists in.

The DynamoFL approach incorporates new federated learning techniques, which address the problem of data governance and privacy by training AI models across multiple decentralized edge devices or servers that store local data samples. The innovative platform also can be used to build the data infrastructure needed to ensure that client devices work together seamlessly and cohesively. It has the potential to handle millions of devices across multiple industry verticals.

We invested in DynamoFL because of its unique capabilities to provide a combination of personalization and performance, without trading-off on either. DynamoFL is a Y Combinator company from the class of 2022. We joined a $4.1 million seed round, led by Nexus Venture Partners. 

The company was founded by two PhD graduates from the Massachusetts Institute of Technology (MIT), Vaikkunth Mugunathan and Christian Lau. They are world-class researchers in the field of federated learning – having published several academic papers. They also have demonstrated their entrepreneurial skills by securing key pilot customers and building strategic partnerships.  

DynamoFL has a unique opportunity because none of its competitors enable personalization on a federated learning platform. Moreover, according to a McKinsey & Company report, companies that excel at personalized machine learning solutions generate 40% more revenue than competitors with one-size-fits-all approaches. DynamoFL models can be personalized using both individual user data and general industry data. Users can quickly protect the privacy of their machine learning and data pipelines using the company’s federated learning module.

The time is right for a personalizable and scalable federated learning platform that can be used across industry verticals. The need for data privacy protection is paramount in this type of environment, and DynamoFL has the right solution at the right time.  

Hina Dixit is an Investor at Samsung Next. Samsung Next's investment strategy is limited to its own views and does not reflect the vision or strategy of any other Samsung business unit, including, but not limited to, Samsung Electronics.

If you’re a founder, we’d like to meet you.

Previous
Previous

Why we invested in Krypton, a decentralized crypto exchange

Next
Next

Why we invested in Lynk, a branded payments platform