Ty Smith
The head of developer infrastructure at Uber who focuses on the scaling pressure that surging code volume places on CI/CD pipelines.
Timeline (last 90 days)
Uber's Ty Smith said that as code volume surges, the build, test and deployment systems are under enormous scaling and efficiency pressure, and the team is catching up with the growth through means such as remote build execution and caching optimization.
Ty Smith believes that as long as engineering systems have safety nets such as feature flags, blue-green deployment and observability, and integrate data such as incident tracking and behavior monitoring into a complete engineering context, even direct PR merges can be kept safe through pipeline guardrails.
Uber adopted a flexible token limit approach: after developers hit the limit, they can submit a justification with one click for manager approval, and Ty Smith emphasized that this not only controls costs but also prompts engineers to think about whether their model choices are reasonable.
Ty Smith cautioned that AI is making all companies more ambitious, but infrastructure scaling must keep up, and audits need to be done in advance to figure out what the next system to crash will be.
All Facts (4)
Uber's Ty Smith said that as code volume surges, the build, test and deployment systems are under enormous scaling and efficiency pressure, and the team is catching up with the growth through means such as remote build execution and caching optimization.
50%UnverifiedTy Smith believes that as long as engineering systems have safety nets such as feature flags, blue-green deployment and observability, and integrate data such as incident tracking and behavior monitoring into a complete engineering context, even direct PR merges can be kept safe through pipeline guardrails.
50%UnverifiedUber adopted a flexible token limit approach: after developers hit the limit, they can submit a justification with one click for manager approval, and Ty Smith emphasized that this not only controls costs but also prompts engineers to think about whether their model choices are reasonable.
50%UnverifiedTy Smith cautioned that AI is making all companies more ambitious, but infrastructure scaling must keep up, and audits need to be done in advance to figure out what the next system to crash will be.
50%