Voxel51

Voxel51 is the computer vision and machine learning company that spun out of the University of Michigan in 2017 and has quietly become a critical piece of infrastructure for the global ML community. The company's open-source product, FiftyOne, is used by researchers and practitioners at every major AI lab to explore, curate, and understand visual datasets. The commercial product, FiftyOne Enterprise, extends that workflow into production ML teams.

The technical work is deep. Computer vision is a data-intensive discipline, and one of the hardest problems in the field is not training a model, but understanding what is actually in the data and what the model is doing with it. Voxel51 built the toolkit that makes that visible. The platform handles dataset versioning, sample-level exploration, model evaluation, embeddings visualization, and the kind of workflow debugging that turns a slow-moving ML project into a fast-moving one.

The company was co-founded by Jason Corso and Brian Moore, both with roots in the U-M EECS department. Voxel51 has raised roughly 50 million dollars and maintains a growing team of engineers, researchers, and developer advocates, with a strong remote presence alongside its Ann Arbor core. The alumni and advisor network runs deep into U-M's CV and robotics community, which matters because serious CV work benefits from proximity to serious CV research.

For ML and computer vision engineers who want to work on the tooling layer of AI rather than the model layer, Voxel51 is a rare opportunity. The mission is clear, the product is used by the people who shape the field, and the Ann Arbor home base offers the full quality-of-life upgrade relative to coastal options. The open-source community around FiftyOne also means the work is highly visible, which is a strong career asset for anyone in the field.