AWS Summit NYC 2026 | David Yanacek, AWS

Clips
More from AWS Summit NYC 2026

David Yanacek

Senior Principal Engineer

AWS

play_circle_outline AWS DevOps Agent: autonomously reacts to production incidents and rolls out changes safely

play_circle_outline Customer case Derive: caught database permission change and memory leak before production

play_circle_outline Automated Release Readiness with Neuro-Symbolic Proofs: Scaling Deployment Validation and Access Control for Coding Agents

Info
Transcript

David Yanacek, AWS

David Yanacek

Senior Principal Engineer AWS

In this interview from AWS Summit NYC 2026, David Yanacek, senior principal engineer at Amazon Web Services, joins theCUBE + NYSE Wired's Gemma Allen to discuss how AWS DevOps Agent is shifting the enterprise bottleneck from code generation to safe, autonomous deployment and operations. Yanacek explains how DevOps Agent autonomously reacts to production incidents, validates code changes before they reach production, and helps organizations keep pace with the surging output from coding agents. He details how the Release Readiness Reviewer uses neuro-symbolic A... Read more

explore Keep Exploring

What are you currently working on at AWS, and what does it do? add

What did you launch today? add

How do the use of many coding agents change the bottleneck in software development, what deployment and integration challenges does that create, and how does a DevOps Agent's Release Readiness Review help address those challenges? add

bolt Powered by CUBE AI

David Yanacek, AWS

search

Gemma Allen

>> Welcome back to theCUBE coverage here at AWS Summit 2026. We're at the Javits Center in New York City, bringing you all of the action on the ground as we talk to some of the folks building this next wave of the AWS ecosystem. Joining me now is a man who is crucial to the development of that. David Yanacek, Senior Principal Engineer at AWS. Welcome, David.

David Yanacek

>> Oh, thanks for having me. It's so great.

Gemma Allen

>> I sometimes run a bio through an LLM to say, "Hey, describe this guy." And I'm going to tell you how ChatGPT described you.

David Yanacek

>> Okay. I'll be excited to hear.

Gemma Allen

>> It said, "He is one of the key engineering brains behind AWS's three and most important frontier agents." So that makes you sound like a pretty important guy, David, I will say.

David Yanacek

>> Oh, I don't know. I guess there's a humility that might not be super baked into that, but I'll take it.

Gemma Allen

>> But you have been key to this event today. I know you mentioned you actually built the demo stage. So maybe talk to me a little bit first about what exactly it is that you really do outside of the ChatGPT description. Bring it to life for me.

David Yanacek

>> Sure. Okay. Well, what I work on currently is AWS DevOps Agent and I'm so excited about that. What it does is autonomously reacts to production incidents. It helps you operate. It helps you peer around corners to help you improve your systems before things happen. And then as of today, it also helps you roll out changes safely that you're making with all your coding agents to get them into production without causing any problems. And DevOps Agent is really kind of the culmination of my last 20 years at AWS doing DevOps. We live and breathe DevOps here, meaning developers do the operations. There is no difference. There is no separate team. And so, that means operations from carrying a pager and getting woken up at 3L00 in the morning, and it also means you need to get your changes out to production. And if they get stuck, you have to unstick it and everything. And so, we gravitate toward what tools will make my life easier? Kind of, very selfishly. We say, "Well, what can I build to make it easier to do all of that operation stuff so that I can focus on the development and actually shipping things for customers?" And so, I've been building and building AWS service after AWS service to just make my own life easier selfishly.

Gemma Allen

>> I love that. Well, 20 years in DevOps, I mean, what a journey too, right, in terms of what DevOps meant then and what it means now, and what it will mean five years from now. Because we all know, this world of tech, there has been a huge shift in so many levels. But it's interesting that you say there that you, I guess, help folks really bring this opportunity to life. Maybe talk us through a typical customer client example, somebody who the before and the after of working with AWS and benefiting from the folks in your organization's great minds.

David Yanacek

>> Great. Okay. So what we launched today is DevOps Agent being able to autonomously test and verify and assess any change that you're making with a coding agent or anything before it goes to production. And so, there's one customer in particular, well, there's several, but there's one customer called Derive. They are a trading platform. And they need to be running 24/7. They're a 24/7 trading platform. So resiliency is very important to them. And so like anybody, they're making a bunch of software changes and infrastructure changes in order to build new things for customers. And they used our release manager capabilities of DevOps Agent, the Release Readiness Reviewer, which it looks at your code, it actually checks it out, it tries it. It uses some neuro-symbolic AI on it. I can talk about that in a bit. But they actually found that it caught several things that would have caused production impact if it had gone to production, a database permissions change that would have broken production and a memory leak that would have over time very slowly messed up production, which those are pretty hard to catch. So they've already had success in catching issues from whoever codes it, whether it's an agent or a person.

Gemma Allen

>> Wow. So this is almost like a sense check of practicality when you move something from, I guess, envision to production. You mentioned when you started talking there that also people want things done faster, easier, right? The expectations are shifting all the time. We see a lot. I hear a lot about Five coding, about all of these new trends in tech. How are you seeing those things, and suddenly everyone's a technologist, play out in your work, in your day-to-day? And what kind of lessons, I guess, are you seeing from the space?

David Yanacek

>> We're seeing developers get a lot done. I myself, instead of working on one thing at a time, I might have half a dozen coding agents all working on a different thing that I need to get done. I just think of a thing and I assign it to a coding agent and it works on it. And so, just the volume of code changes that are happening is just through the roof. And so, what we need now... That has just shifted the bottleneck from the actual act of coding, it's now the act of getting into production and operating that system with all of that rate of change happening. And so we've needed, we've found that changes are just kind of piling up. If there are all these changes happening at once, how do they interact with each other? How do you make sure that none of them individually have a bug in them? Because if anyone... If you have a lot of changes being built all at the same time, you have to deploy them together. So if one of those changes has a problem with it, that kind of spoils the whole batch of apples or whatever. And so, it's really important to help agents that are coding see their own problems, their own mistakes early. That's really all the agents are. Agents are just things that keep trying until they get it right. And so, that means that they need to be told whether or not they got it right. And so, that's where DevOps Agent's Release Readiness Review comes in because it will tell the agent, "Hey, you've made these mistakes and here's the proof." It maybe is, here are the test cases that it figured out and created on the fly to validate the problematic behavior, or it'll use automated reasoning, which is a neuro-symbolic AI, which actually will model the system that the customer has and provide mathematical proofs about the logic being right or wrong around access control changes.

Gemma Allen

>> Wow.

David Yanacek

>> So it's just a coding agent... In order to be correct, it needs to be told whether or not it is correct and that's where the Release Readiness Review comes into play.

Gemma Allen

>> Wow. And let's talk about some of the, I guess, challenges and risks, right? Like security, like ongoing maintenance and monitoring and observability, especially at this time where this idea of agents coding, agents buying, it's all so new. What sorts of concerns or challenges do you see in the space and do you think there's any that are somewhat underestimated by the kind of external world of tech?

David Yanacek

>> I think when it comes to making sure that systems are operated well, it's very important. Everything outside of just coding remains super important, like making sure that things are validated, things in production, that they're working the way that you expect. So it's important to be able to catch issues when they happen, and when they do, to resolve them as quickly as possible. And so, fortunately agents help with that, too, with this kind of after things are in production, how do we make sure that they're operating the way that they want? And that's another part of why we built DevOps Agent, is so that when something does go wrong, that it will root cause that automatically to reduce that time to resolution. We're seeing customers get like three to five X faster time to resolution when they're using DevOps Agent to kind of get that running start. It starts working on figuring out what happened before a human even gets paged. And before their pager goes off and they open their laptop, DevOps Agent is already figuring out what happened and what can somebody do about it. So yeah, we're definitely seeing how agents can really help with more than just coding, with all the other aspects of things that need to go right for any system being secure, being always available, performing well.

Gemma Allen

>> And when we think about the newer opportunities that have arisen as a result of all of this craze that we're seeing right now in the market, especially when we think about the world of DevOps and DevOps Agent in particular, are there any, I guess, surprising customer cases or surprising examples whereby folks might not necessarily have thought about DevOps in that specific way or format like three years ago? Where is the new growth and the new focus coming in?

David Yanacek

>> I think it's just surprising that DevOps and the application of AI to DevOps is something that like a lot of people think about. A lot of people are like, "Okay, yeah, that could save me some time." But I see people surprised when they actually see it in action, when people actually see what it can do and tangibly, "Oh, that is saving me this much time." I think they're just actually... That realization about how good it actually is is pretty surprising. One example is so just last week, so I'm a software developer, I do DevOps, so that means I go on call. Last weekend, I was on call and I got paged. And so a few of us were troubleshooting something together. And so we thought, "Okay, well, we have a..." Anytime you're doing this kind of incident management mode and trying to figure out what can we do to fix this right now, we're always thinking, "Okay, we have a lot of questions and a lot of leads." Well, thankfully, and I've been doing this for 20 years, there's always something going on that we need to figure out and fix fast. And so, this time I was thinking, "Oh, I need to go... We have a lot of questions, a lot of balls in the air, a lot of things to try to figure out. Let's go page in a couple more people." Well, it was the weekend. I would feel a little bit bad bringing people out of their weekend so that they would have to help with this. I was like, "Well, actually, that's the kind of question I could just ask DevOps Agent." It had already started and got us in the right direction. I could just ask these follow-up questions and it could maybe figure it out for us and help us just with the next step of the investigation. And it did. So actually, it made it so I didn't have to bring in more people on the weekend to figure something out. It was actually pretty incredible just thinking about that, yeah, this change is actually here.

Gemma Allen

>> Yeah.

David Yanacek

>> We can actually get way more done now and it just makes our lives so much easier.

Gemma Allen

>> And it can give people some of their time and precious time back, right? Especially last weekend, hopefully they were watching the Knicks.

David Yanacek

>> Right, right, exactly.

Gemma Allen

>> So David, you obviously work across the stock, you see a lot of different opportunities, innovations. Last question to you, what are you most excited about from this summit? And when you think about the world of AWS and we're here a year from now, hopefully, what are you kind of building towards?

David Yanacek

>> I'm most excited about just agent autonomy. That's really where agents, you really want to be able to let them go and just keep working, keep churning. But in order to do that, they need good guardrails. Again, they need to be corrected. They need to be told like, "Okay, now actually go this way. No, you didn't come to the right solution." Or, "That's not the tool I want you to use." And I'm seeing other, just all kinds of things being released today. And just I think we'll see going forward around these things that nudge agents in the right direction from this Strands is an agent building open source framework that we've released. It has now a command line that you can provide to agents so that they have a more constrained set of tools to use to agent core policy to help nudge agents in the right direction. I just think that's where we're going to be seeing a lot of. I'm most excited about agent autonomy. We're seeing a lot of people use agents just like a command line on their laptop. But then when you close your laptop, the agent stops working. So in order to have an agent that's remote, that's actually working all the time, that's autonomous, that's actually a whole different bar of things that need to happen for that agent to be able to trust it to go autonomously. We started there with DevOps Agent. Now we're seeing things like an Amazon Quick autonomous, Curo having autonomous, like the different offline things are working when you don't have your laptop open. And I think that's really where we're going, is like things that are just always helping you, always working.

Gemma Allen

>> It's the claw economy, I guess, as they say, right? Well, David Yanacek, thank you so much for joining us on The Cube. Great conversation. Hopefully see you back here next year.

David Yanacek

>> Great. Thank you for having me.

Gemma Allen

>> I'm Gemma Allen here at TheCUBE at AWS Summit 2026 in the Javits Center. We are bringing you all of the action, all the conversations with AWS execs and some of their key ecosystem players. Stay tuned.