Founder in Focus: Jeremiah Lowin of Prefect

I recently caught up with Jeremiah Lowin, founder of Prefect, an exciting AI startup with offices in Washington, DC and San Francisco. Jeremiah has a Finance/Risk Management background. The company is setting the standard in dataflow automation used to build, run, and monitor millions of data workflows and pipelines.

While his father is a value investor and entrepreneur, Jeremiah likes to dabble in side projects that catch his interest, but having started a business a decade ago, being a founder again wasn’t something he was looking for.

Constantly experimenting, Jeremiah discovered people wanted to pay him for what he was building. He thought it might be small potatoes. Selling for a few dollars, lunch with a friend. Small scale. Another fun side project. The feedback and the appetite for what became Prefect was huge. With expertise in Machine Learning and Data Science, and having worked with diverse stakeholders in large financial organizations, Jeremiah has the ability to deliver automation tools that scale. Under the circumstances, it became clear that founding a company was the best way to take advantage of the opportunity. Moving forward, he says, the need to found was obvious.

Existing use cases at big companies were unimaginative. Query, compile, report. Jeremiah says there was no speed, no scale. He leveraged his research on recurrent neural networks to address dynamic, hard to train, get in trouble fast, projects.

His aim is to save engineering effort, mitigate the cost of being wrong, and minimize opportunity costs. He began to squeeze all the current automation tools to develop the AI tools that are now Prefect. Jeremiah says many people make mistakes when trying to interface with existing tools in AI and modern tech because they aren’t actually practitioners themselves; they can only address what they observe to be problems in other people’s workflows. The tools were big company, big problems; and too many semantics with data. Prefect’s dataflow AI automation received extraordinary reception in the Open Source community. Users can adapt the tools workflows, rather than trying to adapt their workflows to the tools.

With dataflow, 99 percent of the time it works, but when it goes wrong, catching and identifying problems in real time alleviates the unexpected “all hands on deck” hassles and expense of trying to figure out what happened.