Gergely Orosz · 7 min read · AI & Product
Uber’s AI infra, from the inside
Read the original on newsletter.pragmaticengineer.com →
This episode features Uber Distinguished Engineer Uday Kiran Medisetty discussing the evolution of Uber's internal developer infrastructure, from custom-built tools like M3DB and Cadence to modern AI-powered systems like Slate and agentic AI workflows. Uber's unique scale and operational complexity—including cities as a first-class infrastructure concept—forced the company to build most of its tooling in-house. The conversation explores how this pre-existing infrastructure positioned Uber well for the transition to AI agents in software development.
Key Takeaways
- Uber grew from 2 services in 2015 to thousands by 2019, but has since consolidated back to a similar count despite being 10x larger—indicating significant architectural maturity.
- Uber built foundational tools like M3DB (handling 500M metrics/second), Cadence (durable workflow engine), and Morpheus (feature flags + experimentation) because no off-the-shelf solutions could meet their scale or requirements.
- Several open-source companies were founded on Uber's internal tools: Chronosphere built on M3DB, and Temporal built on Cadence.
- Cities are a first-class infrastructure primitive at Uber, influencing observability, feature flagging, deployments, and database sharding—a key driver for custom tooling.
- Uber is now leveraging its mature internal infrastructure to power agentic AI tools like Slate for development and bug-fixing, with a focus on measuring productivity as the engineer's role evolves.
Excerpt
Stream the latest episode Listen and watch now on YouTube , Apple , and Spotify . See the episode transcript at the top of this page, and timestamps for the episode at the bottom. Brought to You by • turbopuffer – The turbopuffer engineering team is doing something really cool: they are completely redesigning their storage architecture from first principles to make search faster, cheaper, and more reliable at scale. And they’re documenting all of it! Follow along their rewrite at turbopuffer.com/v3 • Linear – one of the big challenges for us devs is the sheer amount of code to…