Jeremiah Lowin

Before Prefect, Lowin worked in investment risk and data, joining King Street Capital’s risk team in 2007 and later consulting on machine learning. He completed a master’s degree in statistics while an undergraduate at Harvard. His work automating analytical processes led him to contribute to Apache Airflow, then develop the software that became Prefect. He founded the company in 2018 and remains its CEO.[1][5]

Make MCP servers and clients easier to build

FastMCP provides a Python framework for exposing functions, data and prompts through the Model Context Protocol and for connecting clients to servers. Lowin created the project, and his biography lists it among the tools he built. Its framework handles the surrounding protocol machinery so developers can concentrate on the interface their application needs to expose.[1][2]

Compose an interface as structured data

Prefab lets Python authors express component hierarchies with nested context managers. Reactive variables connect inputs to displayed values. The resulting interface serializes to JSON and renders through React, separating the authoring language from the output representation. Lowin’s introductory article includes live examples, while his conference talk explains why that serializable representation can be produced and modified by either a person or an agent.[3][4]

Operate workflows when things go wrong

Prefect turns Python functions into workflows with execution state, retries, caching and observability. Lowin’s company biography describes the original problem as coordinating data-science work while making failures discoverable and recoverable. The open-source engine lets teams express task dependencies and run workflows locally or through managed deployments.[5][6]