How do agents support the traditional ML development cycle? From exploratory analysis, model protyping, to large-scale deployment & visibility, agentic workflows can accelerate the deployment of traditional ML pipelines with smaller teams. This talk discusses recent work at Duolingo using agent-supported development to test and launch an entirely new set of ML models and infrastructure over the past few months. It will outline our own specific best-practices and pitfalls as well as explore the usefulness of common analogies for AI/developer relationships that frame these practices.
Speaker
Badr Albanna
AI Research Engineer @Duolingo, Neurophysics Professor @Pitt, Formerly @Fordham
Badr is an AI Research Engineer at Duolingo. Currently, he designs and scales bandits and other causal models for monetization Duolingo. Previously, Badr worked on personalized bandits for notifications as well as designing & implimenting novel models for learner knowledge and memory.
Before starting at Duolingo, Badr was faculty at the Unviersity of Pittsburgh and previously Fordham University in NYC. As a professor of neurophysics, Badr researches at the intersection of neuroscience, statistical mechanics, information theory, and machine learning. He is committed to using these technical skills to build products that empower learners in new ways.
