Senior Engineering Manager, AI Platform

The role

This management role oversees a recently created group that builds the internal toolset behind Duolingo's AI features. The team covers dataset management, model evaluation, and AI workflow construction. The position works closely with heads of AI, product, data science, and platform teams.

What you would do

• Set direction for tools that help product teams ship AI-driven functionality quickly and dependably.
• Oversee the shared systems for AI datasets, prompt assets, and evaluation runs, spanning both human reviewers and model-based judges.
• Create repeatable methods that let groups define what success looks like, select suitable models, and tune prompts using feedback loops.
• Simplify routine AI build steps so engineers with mixed levels of AI experience can stand up systems rapidly.
• Partner with the core AI infrastructure group to keep AI pipelines reliable, measurable, and affordable to run.
• Co-develop new shared tooling with leaders from AI, product, data science, and platform organizations.
• Maintain multi-quarter plans and cross-team agreement aimed at faster AI feature delivery and clear product outcomes.

What they asked for

• Prior experience leading and growing a group of software or AI engineers, including hiring and retaining strong people.
• Hands-on record of creating AI-focused infrastructure or customer-facing products powered by AI.
• Working knowledge of AI assessment techniques, particularly blends of human review and LLM-based judging.
• Demonstrated skill leading across functions and explaining plans to AI, product, data science, and platform counterparts.
• Comfort steering through unclear situations and defining better practices in a rapidly changing AI field.
• Completed a bachelor-level degree in computing or an adjacent technical field.
• Able to be based in Pittsburgh or New York, either already or after relocating.

Nice to have

• Deep involvement in building evaluation frameworks and data workflows that keep people in the decision loop, at large volume.
• Background building internal tools for a platform group, exposed through approachable APIs, protocol servers, or web interfaces.
• Past decisions weighing in-house construction against purchasing, plus shaping long-range plans around internal user needs.