Eiso Kant

Kant began studying at IE University in 2009 and co-founded the recruiting company Tyba. His own 2015 essay describes building software to help people find work they enjoy. Poolside’s later founder account describes his work on source{d}, an earlier company applying machine learning to code. His current Axon biography also identifies his work developing AI compute through Poolside Infrastructure Company.[3][4][6][2]

In an article coauthored with Joe Rowell, Kant describes how Poolside runs code during model training. Repositories require reproducible environments, dependencies and isolated execution; a text-only model response is insufficient. Container images and shared layers reduce the cost of recreating related environments, while revision-aware serving makes the required environment available to workers. This infrastructure lets tests and compiler results feed back into training.[5]

Kant’s execution article makes infrastructure part of the learning problem. Slow environment setup means fewer experiments and longer waits for feedback. Improving that loop requires reliable execution as well as a stronger model; the article describes an engineering approach rather than a claim that every generated solution passes its tests.[5]