Ryan Stevens

Ramp

BIO

I am currently leading the Core Risk and Core Growth Applied Science and Analytics Engineer teams at Ramp. Our teams build the core ML models and data products powering Risk Management and Growth at Ramp. Prior to joining Ramp, I worked as a Data Scientist at Meta on their Ads Core ML team. Before that, I completed my Economics PhD at New York University. My academic research interests were primarily focused on industrial organization, applied econometrics and behavioral economics. Outside of work, I live in Asheville, NC with my wife and our pup Archie. I like to code, play around with interesting datasets, and bike around.

TITLE

From Noisy Alerts to Smart Fixes: AI in Incident Response

ABSTRACT

AI is now on-call. At Ramp, we’ve built systems that parse logs, identify root causes, and suggest fixes—reducing stress, wasted time, and late-night panic. But the real story isn’t just the system we built—it’s why we chose this problem.

Too many AI projects fail because they target jobs instead of tasks. We approached incident response not as a monolith, but as a series of narrow, well-defined tasks: gather logs, localize the error, and suggest a fix. This task-based framing made the problem tractable for AI and the solution impactful for engineers.

In this talk, I’ll walk through the AI system we built. Along the way, I’ll talk about a task-based framework to consider the ideal candidates for AI solutions.