Willem Pienaar is an engineering lead at Tecton and the creator of Feast, an open source feature store for machine learning. His career has focused on one of the hardest problems in production ML: getting features to serve reliably in real time without duplication, rework, or skew between training and serving.

At Tecton, Willem works on the feature platform that grew out of his open source work. He has written and spoken extensively on where feature stores fit in modern MLOps stacks, covering architecture components like transform engines, storage, serving layers, and registries, along with integrations with tools such as dbt, Kubeflow, Airflow, Flink, and Spark. A favorite example from his talks is real-time fraud detection, where low-latency feature lookups can make or break the system.

As Feast’s creator, Willem has helped shape how machine learning teams think about feature management, governance, and reuse across projects. Teams that want to explore Feast and feature stores in practice can start with the resources at feast.dev.