This talk will explore the role of evaluations (evals) in AI agent development, from full-agent assessments to micro-evals. Drawing parallels to traditional software engineering, the talk will discuss why evals are essential for building reliable AI agents that actually solve customer needs and lessons learned from shipping agents at Datadog.
This talk will explore the role of evaluations (evals) in AI agent development, from full-agent assessments to micro-evals. Drawing parallels to traditional software engineering, the talk will discuss why evals are essential for building reliable AI agents that actually solve customer needs and lessons learned from shipping agents at Datadog.
Evan has more than 10 years of experience building customer-facing distributed systems, founded an AI startup in the image generation space, and now hacks on code generation agents at Datadog.
Evan has more than 10 years of experience building customer-facing distributed systems, founded an AI startup in the image generation space, and now hacks on code generation agents at Datadog.
Joey is a born and bred New Yorker with an MEng in Computer Science from Cornell Tech. He built low latency distributed systems that process petabaytes of data per day at Datadog and now works on building AI agents at Datadog.
Joey is a born and bred New Yorker with an MEng in Computer Science from Cornell Tech. He built low latency distributed systems that process petabaytes of data per day at Datadog and now works on building AI agents at Datadog.
