# Building Blocks for Uber’s Software Factory— Uday Kiran Medisetty & Adam Huda, Uber AI Engineer · 2026-08-21 Uber's Uday Kiran Medisetty and Adam Huda explain the six building blocks of their agentic software factory, which now generates more than 70% of pull requests and has doubled lines of code per engineer year over year. A model gateway handles 100 million requests a day across 800 projects with redaction of 20+ PII types and safety models under 100 milliseconds; an MCP gateway cuts token use over 40%, a skills marketplace runs 20,000 executions daily across 2,500 skills, and a context graph of 40 million entries replaces 20-30 systems. Adam shows one feature end-to-end, from Slack idea to draft PR that stops before CI, validating via simulator screenshots against Figma in an inner loop. The bottleneck, they argue, has moved to deciding whether a thing should be built at all. ## Questions this episode answers ### What percentage of Uber's pull requests are now generated by AI agents? Uday Kiran Medisetty says over the last year, agentic AI led to more than 70% of Uber's PRs now either by local or cloud agents, and this resulted in twice the number of lines of code per engineer year over year, extending beyond coding across the engineering lifecycle. [0:34](https://aiengineer.podhood.com/5737561a-6452-4021-82fe-ac1d5ee61a7f?t=34000) ### What is Uber's model gateway and what guardrail budget does it have? Uday Kiran Medisetty explains Uber's model gateway is one OpenAI/Anthropic-compatible endpoint used by all internal and external use cases, with Spire identity, a data anonymizer redacting 20 plus PII types, and five specialized safety models, all running under 100 milliseconds. It handles more than 100 million model requests per day across 800 plus projects. [1:46](https://aiengineer.podhood.com/5737561a-6452-4021-82fe-ac1d5ee61a7f?t=106000) ### How did Uber reduce token consumption by more than 40%? Uday Kiran Medisetty describes token optimization strategies in Uber's MCP gateway: Omni MCP, projecting MCPs into a CLI pattern so responses stay out of context, and a code mode skill that builds Python scripts to hyper-optimize top MCP token consumers. Together these delivered more than 40% fleetwide savings. [4:45](https://aiengineer.podhood.com/5737561a-6452-4021-82fe-ac1d5ee61a7f?t=285000) ### What is Minion and why does it stop before CI? Adam Huda says Uber's Minion agent is a cloud coding agent powered by dev pods, working across repos in interactive or autonomous mode. It stops at creating a draft PR and does not push to CI yet, so validation can happen earlier in the inner loop and avoid extra load on CI. [13:18](https://aiengineer.podhood.com/5737561a-6452-4021-82fe-ac1d5ee61a7f?t=798000) ## Key moments - **[0:00] The numbers** - [0:49] Uber's agentic AI shift: 70%+ of PRs from local or cloud agents doubling lines of code per engineer year over year — Uday Kiran Medisetty - **[1:26] Model gateway** - [3:23] Uber's model gateway: Spire identity, 20+ PII redaction, five safety models under 100ms, 100M requests/day across 800 projects — Uday Kiran Medisetty - **[3:51] MCP gateway** - [5:28] Uber's MCP gateway hosts 1,000+ MCP tools and cut fleetwide token use by 40%+ via CLI projection — Uday Kiran Medisetty - **[5:48] Dev pods** - [6:32] Uber agentified dev pods into pre-provisioned Kubernetes balloon pods with snapshotted repos and built search indexes — Uday Kiran Medisetty - **[7:18] Skills marketplace** - [8:24] Uber's managed skills marketplace holds 2,500 skills behind lint checks and automated reviews, running 20,000 executions/day — Uday Kiran Medisetty - **[8:39] Context graph** - [9:32] Uber's context graph of 150 node/edge types and 40M entries replaces 20-30 scattered systems for ownership facts — Uday Kiran Medisetty - **[10:20] Cortana** - [11:19] Uber's Cortana assistant hit 300 unique personas created and 20,000 sessions per day in the last month — Uday Kiran Medisetty - **[11:35] Building a feature** - [12:42] Adam Huda shows a World Cup pickup idea turned into Figma A/B variants via Cortana, compressing weeks into minutes - [13:47] Uber's Minion agent stops at a draft PR before CI so inner-loop validation compares simulator screenshots to Figma specs — Adam Huda - **[14:18] Inner loop** - [15:45] Autonomous Minion PRs include a table of self-improvement checks and simulator screenshots for human reviewer confidence — Adam Huda - **[16:10] Maintenance** - [16:25] Uber's managed maintenance skills loop Sundays for feature-flag cleanup and cap Monday diffs to avoid overwhelming engineers — Adam Huda - **[17:32] Bottlenecks** - [17:44] "It's not about, you know, can we build? We know we can probably build it now. It's more of a question of should we build it?" — Adam Huda ## Speakers - **Adam Huda** (guest) - **Uday Kiran Medisetty** (guest) ## Topics Coding Agents, Agent Engineering, Developer Productivity ## Mentioned Google (company), Uber (company), Bazel (product), Cortana (product), Figma (product), Jira (product), MCP (product), Minion (product), Slack (product), Spire (product) ## Transcript ### The numbers **Uday Kiran Medisetty** [0:13] Hey, uh, let's get started. Good morning, everyone. I'm Uday, I'm here with my colleague Adam. We'll talk about our journey toward Managed Software Factory. And in the beginning, in the first part of the talk, I'll talk about the key building blocks that we're investing in, and later Adam's going to talk about how we take all of these blocks to build an end-to-end cohesive solution for our engineers. To set some context, we have a few thousand engineers across 12 global tech sites. Over the last year, all of the investments we made in agentic AI have led to more than 70% of our PRs now either by local or cloud agents. And all of this led to twice the number of lines of code per engineer year over year. And this extends way beyond coding, and we see it in every aspect of the engineering lifecycle today. And we're also accelerating toil at a, toil reduction at a massive pace. We handled more than 250 automated migrations, cumulatively 9 million lines of code automatically for our engineers. And before even the building blocks, um, you know, all the investments we made over the last 6 years on moving to monorepos, moving to Bazel, all of that also laid a really solid foundation for us to accelerate this. ### Model gateway **Uday Kiran Medisetty** [1:27] So the first, I'll cover all of these six building blocks, and Adam's going to talk about a specific example and show how that feature can be built end-to-end, uh, with all of these. And all of these are in various stages of maturity and rollout within Uber, but we want to give everyone a sneak peek of what we're up to. So let's go to the building blocks, the six building blocks, one by one. The first one is model gateway. This is one of our earlier investments. The three things that we wanted to make sure, ensure was no PII ever leaves our perimeter to any of the vendor by default. And any guardrail that we add here, the latency of that is strictly bounded. And every request that goes through this, whether it's, uh, we need to be able to attribute per user, per project, and per team. So we have a model gateway, and we made sure all of our internal use cases, our coding hardnesses, our external use cases, they all go through one single OpenAI Anthropic-compatible endpoint. It goes through a series of middlewares. The first one is identity and authentication using Spire. Uh, we have a data anonymizer that redacts 20 plus PII types. We have an AI guard that has five specialized models that handles various parts of safety and policy that we want to ensure, and all of that runs under 100 milliseconds. We also are investing in all kinds of caching and token optimization strategies at this layer. And every request that goes through this, we are able to attribute to a specific project in our catalog, and we can attribute per caller, per user, per team, both in real time but also in our data lake. This enables us to create all kinds of spend tiers and guardrails in a holistic way across our portfolio. We also use this layer for capturing audit logs, session traces, which are then plugged into our benchmarking and all kinds of self-improvement loop efforts. And for an engineer at Uber, you take the vanilla client, you set the project ID, and we, and we take care of everything else. Today we have 800 plus projects internally going through this, cumulatively handling more than 100 million model requests per day. This includes both the frontier models and also open-source models, whether that is hosted in our infrastructure or some of our vendors. The next is how do we provide tools to all of these models. Last year when we started on this journey, we had thousands of internal APIs, but none of them are agent taxable out of the box. And we had so many other SaaS tools, and each one of them has a different way to authenticate, different way to set up, which is a lot of hassle for, for everyone. ### MCP gateway **Uday Kiran Medisetty** [4:07] And once you end up with enough MCPs, they'll all add up to and, and have a massive token tax. Similar to model gateway, we have an MCP gateway that handles a whole bunch of middlewares for, for engineers. And we have an automated crawler that looks at our internal APIs and projects all of these into MCPs with one single config change. And we do the same thing even for our SaaS MCPs. Whether it's Google, Slack, Jira, all of this, they go through the MCP gateway. We ha we host them, we do the token exchange. So for all the engineers, they go through one single entry point, one common way to install any MCPs. This simplified a lot for all of our engineers and employees. And then the whole bunch of token optimization strategies. We initially had direct MCP pattern. Earlier this year, we created Omni MCP, which is one single MCP that you install, which can discover and invoke any MCPs within the gateway. And a couple of months ago, we projected all of these MCPs into CLI pattern so that even the response doesn't eat up in your context. And of late, we also have a, a code mode skill, which is auto-installed, which on the fly creates Python scripts to hyper-optimize some of the top MCP token consumer, consuming use cases. And all of this led to, like, now we have 1,000 plus MCP tools, and, uh, just with these optimization efforts, we've saved more than 40% fleetwide savings. So once we have the models and the tools, we need a place to run all of this. For many years, we had dev pod, which is our cloud, uh, remote environments. We, we, we had this because we had, like, large monorepos with millions of lines of code, and this is how engineers work at Uber. ### Dev pods **Uday Kiran Medisetty** [5:58] And now we took what we had with dev pods and we a-agentified that. Now we need some environment for agents to run for a longer period of time. They need to be quick, they need to be isolated, we can install any number of them, and they need to be globally available across all of our sites. So we have a pre-provisioned Kubernetes balloon pods. When an agent requires a new environment to run, it can take one of that, which is already pre-provisioned. It has all of the repositories already snapshotted, the search indexes already built, so the agents can start working within a matter of seconds. The next thing we noticed is the, the, the roles of engineers are getting blurred. We used to offer a dev pod per language flavor for Go, Java, Android, and so on. Now we need agents to work across repositories and engineers also to work across repositories. So we have a mega dev pod that has all of the repositories in one, one common place, and this is what we use for our autonomous coding agents now. And even for our non-engineer employees, we are providing a simple way for them to get started with any of the agent hardnesses in a matter of seconds. Then now we get to knowledge part of it. And we jumped on this bandwagon earlier this year. Uh, we started noticing engineers building tons of skills across many repositories. Um, one three problems we noticed was there's a lot of duplication, same skill being built by different engineers in different repos. ### Skills marketplace **Uday Kiran Medisetty** [7:27] And discovery and configuration was a huge hassle, and a lot of skills were of subpar quality. So what we built an entire lifecycle around skills. So we have core s core skills and domain-specific skills. All of that go into a managed skills marketplace. We have 2,500 skills thereright now. Um, and it goes through a whole bunch of lint checks, automated reviews, which ensures a baseline skill quality for any skills that we have. And we also simplified the installation and discovery. So there is one single command to discover and install any plugin in our ecosystem. And based on the engineer personas, we even auto-install some of the default skills so the agents automatically can pick up theright skill. You don't even have to even install them. And of late, we started working on collecting traces and comments and capturing continuous evals so that we can give feedback back to the skilled authors for skill improvements. And this is an area of big investment for usright now. And we have 2,500 skills and cumulatively more than 20,000 skill executions per day across our fleet. The next piece of knowledge is context graphs. Uh, we st we started noticing in our execution traces agents spending a lot of time even trying to find basic context, especially in our large monorepos. You need to identify where the service is located, what are the dependencies, um, who owns it, what kind of patterns I need to follow. ### Context graph **Uday Kiran Medisetty** [8:56] And all of this context is gathered across scattered s-systems across Uber. There's 20 to 30 different systems. Each needs its own skill, skills, its own MCPs to, to gather the context. And this burns tokens, this adds a lot of latency, and it creates more unpredictable outcomes. So we have one context graph. We took all of the information of how Uber runs into one context graph. This has 150 unique node and edge types. We have 40 million entries thereright now. It captures all the way from how our mobile apps are built to our backend to our data lake. All the design docs, Jira, incident bugs, everything is connected. And this enables agents to quickly find theright context within our ecosystem. We are now plugging all of our skills and use cases into the graph, whether it's our on-call RCAs, whether it's our planning or data analysis or security scans. And we see across all of this, they we are improving, uh, the skills by a lot. And I'm just showing a very simple example of asking a simple question of how many mobility trips in India are, are cached. This needs to understand the concepts of each of these, which tables, what kind of, uh, CTEs you need to create for the SQL. With and without graph, we see massive improvement in tokens, turns, and latency. And we see that across any early eval that we did, uh, within our infrastructure. And the last thing is how do we package all of this for everyone in the company to use. So we have, uh, our AI assistant called Cortana. All of the things that I mentioned so far, whether it's skills, MCPs, and context graph, they're all plugged into that in every surface possible, whether it's on Slack, CLI, web. ### Cortana **Uday Kiran Medisetty** [10:37] So anyone in the company, they can ask a simple question. It can look up the context graph, invoke any skill, check any code, check any code in any code base, and give an answer across any of these surfaces. And now we started allowing employees to even personalize that. You can hook up your custom skills, custom prompt, and hook it up into your team's Slack channel so that it, it knows all of the things about that team and works like a, the t that teammate. And this, this is a simple example of how you can invoke the same question before in Slack. Um, and ma all of the employ m-m, more, like, one or more people can even collaborate on the same Slack channel. And we have, in the just in the last one month, 300 unique personas created and more than 20,000 sessions per day. I'll now pass on to Adam who will talk about how we take all of this and build, uh, take and ship a feature end-to-end. **Adam Huda** [11:34] Allright. Thank you, Uday. Allright. As Uday said, we've got those building blocks. We're going to use those to power our software factory. So we're going to take a feature here and show it going end-to-end through this. Allright. First up,right, we need to have an idea,right? ### Building a feature **Adam Huda** [11:49] A good idea probably for this moment would be something around the World Cup,right? Would it be awesome if you were a writer and you were leaving a busy stadium if there was a better pickup location to get you away from the crowd? So that's the idea,right? We could have our idea, we're jamming on it in Slack here. Let's tag in Cortana,right? That's our AI assistant to help us with that idea. Cortana with that context graph can help us determine whether this is a good business opportunity to go after. So we can go here from Slack and now open Cortana into a web interface. And you'll see an example here of what that business research could look like,right? What other large-scale venue events have happened before? What are some stadiums that would make sense here? From there, we start to think about the product requirements,right? This should be probably just a North America rollout since that's where the stadiums are. We can even then bring in Cortana to help us think about, uh, the Figma designs. We can create some initial mockups,right? Do two variants here. We want to run an experiment, A and B. So the, the button strings here are different between the two. So we'll test those two variants and see which one performs better. Now we'll start to think a little bit about the design, and Cortana too can help us think about what code changes we need to happen,right? What can we leverage that's in the app already? What screens? And what can we leverage on the backend? Right? So this process before could take a long time. It could take weeks to get everyone aligned. Now we can compress this into a very short amount of time now,right? And get to a prototype here very quickly. Allright. So now we got to go build this. So we, we hand off from that Cortana agent to what we have at Uber. We have a Minion agent. It's a Uber's cloud coding agent solution. Allright. So you can use Minion in an interactive mode or you can run it in an autonomous mode as well. So I'm going to show you what this looks like. Um, Uday mentioned the dev pod building block. So this is powered by that dev pod. So it's got a full build environment and it can work across repos. So we're doing backend changes and the frontend changes here too as well. We're going to see Minion kind of progress here, and it's going to stop at just creating a draft PR. It's not going to push it to CI yet. The reason being is that we were seeing, um, uh, uh, that this is great for doing like TOIL sort of workloads, but to build more advanced, like end-to-end features, we really need to be able to validate, uh, the feature first. And we want to prevent, uh, a lot of extra load coming onto CI. So if we can validate, um, sooner before we push to CI, um, that would be a big benefit. So that's what we're going to see here next on validation,right? In the SDLC, we have an inner loop. We have the outer loop, of course. We can have these be agentified where we're shifting more checks now to happen in this inner loop. Allright. So some of the checks that initially happened that we've had there previously is like the, the static analysis sort of checks. ### Inner loop **Adam Huda** [14:33] When those are detected, now we fix those. But we can shift things that happen in inner loop, things like visual validation. So we can launch, uh, a simulator with a skill, grab a screenshot from the simulator, compare it to the Figma specs. We can also bring up the service and our backend staging environment and compare the frontend and the backend integration together. So now that we've moved a, um, now that we've done, uh, that part, uh, we moved to the outer loop where CI, uh, typically happens. Errors can still happen on CI,right? Uh, so self-healing CI is something that we've implemented here where we can fix a lot of the issues that you hit on CI. Uh, code review is another thing that happens in the outer loop, but this is another thing that we've shifted. We've moved parts of code review to happen in the inner loop,right? The outer loop code review can have a powerful model, use reasoning, a skill to do a deeper review. And then the inner loop, we can have a smaller model that runs, uh, faster with a, with a, with a medium model. Now, another key thing here,right, is this, if this is an autonomous diff coming from Minion, we want to give a human reviewer some confidence that this diff has gone through a lot of self-improvement already,right? That does not just touching that initial generation that happened, but all these other steps have happened. And so on the PR, you will have a table attached that says all these different checks that it went through, including the screenshots. Allright. So we've got a lot more code coming through the software factory now,right? Let's talk about maintenance,right? Maintenance is even more important. Um, what we have set up now is we can actually enroll our feature or skill into, um, our feature or serv, uh, service into maintenance, uh, skills. ### Maintenance **Adam Huda** [16:10] So these are, uh, some examples of those skills that we have. Uh, feature flag cleanup,right? We had two variants of that World Cup, uh, modal. Now that the, the B variant is no longer needed, we can have that scaled on a loop. So the key thing here is that this is actually a managed loop that you go to,right? We don't want thousands of loops being set, uh, set up across the company without any bounds. You have a managed surface that you go to to set up the loop so it runs on Sunday when we know we have better, uh, CI capacity available. Uh, we also don't want to overwhelm engineers that Monday morning with a bunch of extra diffs. We want to control how many diffs they're seeing on Monday as well. Another cool key thing here is that when that ski, uh, skill runs and makes those diffs, those diffs will get comments and either get landed or not landed. That's all good label data that we can use to improve the skill itself. And then at a kind of monthly cadence, we're looking to see what skills can we now learn, um, from in our incident reviews and turn those into new maintenance skills that we can apply to all of our services. Allright. So you've seen, uh, these parts of the SDLC that we have agentified. There's other parts too, like monitoring. Have you seen the, uh, building blocks that you can use to power those and the architecture underneath them? One of the other things that we're really thinking about now is bottlenecks,right? We're now, we're putting more strain on our infrastructure. So we're trying to anticipate where our CI capacity needs to be and make theright foundational investments there. There's only so many experiments that we can feasibly run as well. So that's another bottleneck. ### Bottlenecks **Adam Huda** [17:44] And then lastly too,right, decision making,right? It's not about, you know, can we build? We know we can probably build it now. It's more of a question of should we build it? Allright. So that's what we have for you today. And the next talk is going to be actually from Uber as well. And if you want to learn more about our agentic code review, uh, Amea and Will will be presenting that next, uh, in this room. Thank you. --- This library is powered by PodHood (https://podhood.com), the podcast website platform.