Data agents
Autoscience: How we built agent-driven ML exploration at Faire
Our short summary of Faire’s engineering write-up. The full story, with all the detail, is on their blog.
Read the full write-up on craft.faire.com (opens in a new tab)Problem
ML teams could only try a tiny fraction of possible model changes, limiting performance.
Solution
Built Autoscience, an agent system that proposes, runs, and evaluates model experiments with persistent memory and steering hooks, enabling many more iterations.
Production model accuracy improved double‑digit percentages
Want the architecture, code and numbers? They’re in the original post.
Read the full write-up on craft.faire.com (opens in a new tab)