Claypot AI

Platform for Real-time Machine Learning

Streaming technologies are changing the data landscape and every application that produces and consumes data. Yet, most machine learning models, whose performance is tightly coupled with data quality and data freshness, are still in the batch paradigm.

Claypot unifies streaming and batch systems to make it easier and cheaper for companies to do online prediction, continuous evaluation, and continual learning.

Our platform was designed with streaming and ML best practices, with learnings from leading the streaming data platform team that serves over 2000 data use cases at Netflix and helping companies of various sizes develop ML applications at NVIDIA and Snorkel AI.

Our solution can be especially helpful for problems in fast changing environments such as recommender systems, e-commerce, fintech, and logistics.

Claypot was founded by Zhenzhong Xu and Chip Huyen (who also teaches Stanford's CS 329S: Machine Learning Systems Design). We're well-funded and backed by the founders of some of our favorite data/ML tools.

For more discussion on the problem we're tackling, see Machine learning is going real-time and Real-time machine learning: challenges and solutions.

👁 Request a demo
💻 Work with us
📜 Real-time ML survey