Activity
998 followers
-
Ainsley Escorce-Jones posted thisAfter 8.5 years of working on Developer Productivity at Stripe (and hopefully making the developers more productive) I have decided that the next step for my personal development is to become deeply unproductive for a while. No plans in the short term but extremely excited to take a break before deciding what's next. A massive thank you to Scott MacVicar, Yvette Nameth and everyone else I had a chance to work with who made the time at Stripe as fun as it was!
-
Ainsley Escorce-Jones reposted thisAinsley Escorce-Jones reposted thisEarlier this year, Anupam Upadhyay and I set ourselves a new goal: 10x every Stripe in the same way coding agents had 10x-ed engineers. That idea evolved into Stripe’s Knowledge AI Platform, our in-house AI tool. Today, 80% of Stripes use it weekly across more than 20,000 sessions every day. It helps people turn fragmented context spread across various internal systems into answers and usable artifacts. And the impact is being felt all around us - we just learned that Account Execs close almost 40% more deals when they use it! Read how we built a platform that scales expertise across Stripe: https://lnkd.in/g-RV3pcn
-
Ainsley Escorce-Jones reposted thisAinsley Escorce-Jones reposted thisOur on-call engineers used to get paged hundreds of times a year for degraded MongoDB shards, and each incident could take up to 3 hours of manual recovery work. Stripe processes $1.9 trillion in annual payments. Every second matters. We used graph search to fix it. Pathfinding algorithms can look at a broken shard, figure out the fastest route back to health, and just auto-remediate it. 2,000+ MongoDB shards now heal themselves with 30% fewer pages, and 12 days of fleet health reclaimed every year. We also reclaimed hundreds of engineering hours to work on more strategic problems. Learn about how we built it in our new blog post https://lnkd.in/gAr2BXfi #Stripe #MongoDB #DistributedSystems #Automation #GraphSearch #PlatformEngineering #EngineeringHow Stripe uses graph search and state machines to auto-remediate a global database fleetHow Stripe uses graph search and state machines to auto-remediate a global database fleet
-
Ainsley Escorce-Jones reposted thisAinsley Escorce-Jones reposted thisIs each bug a random one-off, or is there a pattern to the bugs AI generates? We clustered 1,000 bugs from 99 codebases to find out. We grouped them by failure mechanism, so the patterns reflect how code breaks, not who wrote it or what language they used. We found that just 21 recurring mechanisms explained 70% of the bugs. These same mistakes show up again and again in completely unrelated products. Agents are creating the same bugs over and over! Awesome analysis by Peter Pyry Takala. Write-up with the full breakdown in the comments. Let us know if you have followup analyses you're curious about. We've found hundreds of thousands of bugs and are excited to use this dataset to better understand the fundamental drivers of software quality.
-
Ainsley Escorce-Jones reposted thisAinsley Escorce-Jones reposted thisI recently wrote about what I've been working on the past two years, and others have built over the past decade: the infrastructure that keeps Stripe's entire developer ecosystem in sync with every API change we ship. We built an OpenAPI-based pipeline that flows API changes from their initial definitions all the way to our SDKs, docs, and other developer products in lockstep, letting engineers focus on their product instead of manually propagating their changes to every developer product that Stripe offers. Read more about how it works under the hood → https://lnkd.in/dRJmwRVxHow API changes flow into Stripe's developer productsHow API changes flow into Stripe's developer products
-
Ainsley Escorce-Jones reposted thisAinsley Escorce-Jones reposted thisThere is a (facetious) saying within Developer Infrastructure at Stripe: “Product teams look at code as pets; DevProd looks at code like cattle.” I’m honored to feature a project I led recently on our external engineering blog that demonstrates this in practice!Modern Java at Stripe: Language upgrades as a serviceModern Java at Stripe: Language upgrades as a service
-
Ainsley Escorce-Jones reposted thisAinsley Escorce-Jones reposted thisLast week, GovTech Singapore visited our San Francisco office to spend time with the engineering team at Stripe. We got into how we’re approaching AI engineering, our work on Minions, developer productivity tooling and how we’re thinking about benchmarking AI-powered integrations. Not the usual conversation around payments, money movement, or revenue growth, but a useful reminder that while payments are what we’re known for, much of the differentiation sits behind the scenes in how the engineering teams build and operate. That’s what allows Stripe to move quickly (288 launches at Stripe Sessions!) while still maintaining five nines of reliability. Thanks to Ainsley Escorce-Jones, Kevin Ho, Carol L. for sharing and Pranav Ashok and Andreessen Horowitz for bringing this together. For those curious, you can read more in the miniseries on our engineering blog here: https://lnkd.in/g3v4GS9j https://lnkd.in/gPFx7PDs #AI #Engineering #GovTech #Singapore #Stripe #SEA
-
Ainsley Escorce-Jones reposted thisAinsley Escorce-Jones reposted thisStripe's Ruby codebase is massive at 50 million lines, backed by roughly 100,000 test files and over 1.2 million test units. The challenge lies in handling 50,000 builds a week (with AI agent-coding increasing this number rapidly) without slowing down developers or sacrificing reliability. Instead of running every test on every build, we built a Selective Test Execution (STE) system to dynamically observe and record exactly which files a test touches. This allows us to safely execute only about 5% of our test suite per build. I blogged about this here: https://lnkd.in/gFJ-wWqQSelective Test Execution at Stripe: Fast CI for a 50M-line Ruby monorepoSelective Test Execution at Stripe: Fast CI for a 50M-line Ruby monorepo
-
Ainsley Escorce-Jones reposted thisAinsley Escorce-Jones reposted thisHey folks! I'm hiring for a very exciting role! I'm looking for a scrappy, extremely product driven EM to lead the DevProd AI team. This job is so cool and we've built an amazing team to massively accelerate Stripe's productivity with AI tooling. One of the things we work on: Minions Blog Pt 1: https://lnkd.in/ed3Tfktk Minions Blog Pt 2: https://lnkd.in/e2hfdKku https://lnkd.in/effGDYf4
-
Ainsley Escorce-Jones liked thisAinsley Escorce-Jones liked thisEarlier this year, Anupam Upadhyay and I set ourselves a new goal: 10x every Stripe in the same way coding agents had 10x-ed engineers. That idea evolved into Stripe’s Knowledge AI Platform, our in-house AI tool. Today, 80% of Stripes use it weekly across more than 20,000 sessions every day. It helps people turn fragmented context spread across various internal systems into answers and usable artifacts. And the impact is being felt all around us - we just learned that Account Execs close almost 40% more deals when they use it! Read how we built a platform that scales expertise across Stripe: https://lnkd.in/g-RV3pcn
-
Ainsley Escorce-Jones liked thisAinsley Escorce-Jones liked thisWe’re announcing our $13.5M Series A led by Standard Capital. AI has taken over engineering and companies are wasting millions. Weave gets you the most out of every dollar spent on tokens. 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 - we use LLMs and ML to show you exactly how much work is getting done and how good the code is. 𝗧𝗼𝗸𝗲𝗻 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 - we pull in every AI tool and analyze everything from prompt to production. 𝗣𝗿𝗼𝗺𝗽𝘁 𝗿𝗼𝘂𝘁𝗲𝗿 - we take this data, look at every prompt, classify it, then route to the cheapest model (without ever sacrificing speed and quality). And it learns over time. If for no other reason, have fun watching me conquer my fears with my acting debut and jumping out of a plane. If your company is trying to figure out AI spend, hit us up. We'd love to help. Huge thank you for the continued support from our early customers, angel investors and Burst Capital, Moonfire, Irregular Expressions, Agent Fund, Pioneer Fund.
-
Ainsley Escorce-Jones reacted on thisAinsley Escorce-Jones reacted on thisI have never once regretted getting on a plane to visit a customer. You know the reasons by now. - You get a deeper understanding of your users and how they work. - You learn what their real problems are – the stuff that doesn't come up over zoom. - You build real relationships that enable higher trust collaboration down the road. All that good stuff. But there's another important reason nobody is talking about. We need to start talking about it. Visiting customers gets you to Waffle House country.
Experience & Education
Patents
Honors & Awards
-
1st Place - Facebook London Hackathon
Facebook
-
1st Place - UCL FIndS Hackin' The City
UCL FIndS
-
Best in Show (SAP Young rewired State 2012)
-
Explore more posts
-
Jason Mattiace
Jensen said what most leaders are still hesitant to say: “If that $500k engineer did not consume at least $250k worth of tokens, I’m going to be deeply alarmed.” He's saying tokens are not just software spend. They are capability spend. They buy faster iteration. More parallel exploration. More prototypes. More test coverage. More research. More options per engineer, per day. The wrong question is: “How do we control token cost?” The better question is: “How much more output, learning, and product velocity do we get per dollar of token spend?” That is why the best teams will not manage AI like SaaS. They will manage it like capital investment. Not every company should spend $250k per engineer. But every company should understand the principle. If your best engineers are still working at human-only throughput, your org design is already obsolete. The foreseeable future is not engineers replaced by AI. It is engineers amplified by so much AI that the old productivity baselines stop mattering.
27
5 Comments -
Ido Shamun
Engineers love protecting code more than shipping product. We gatekeep with complexity. With "proper" processes. With code reviews that take longer than the code took to write. We tell ourselves it's about quality. It's actually about control. Here's what happened when I stopped: I set up a boilerplate project for our designer. Taught him Cursor and basic Git. Then got out of his way. Within days he shipped a website more beautiful and interactive than any developer could have built. Faster too. The guy we were "protecting" the codebase from? He delivered more than we ever did. Every pull request you block. Every "let me just do it myself." Every "you wouldn't understand the codebase." That's not quality control. That's a bottleneck with an ego. Your codebase doesn't need protection. Your business needs results. What are you actually protecting: code quality or your territory?
35
34 Comments -
Ellis Shaw-Simpson
Why early-stage founders are missing out on top engineers 👇 A lot of early-stage U.S. founders think hiring a senior engineer means competing with Google salaries. Not always true. The best engineers I talk to want impact, autonomy, and speed, not just comp. What turns them off? 6 rounds of interviews No clarity on the product roadmap Founders who can’t articulate what success looks like Great engineers join missions, not job ads. You don’t need to outpay the market — you need to outcommunicate it. #TechHiring #AIRecruitment #StartupHiring #RecruitingTips
5
-
Rami Banna
It's a wild time to be building infrastructure. Today we shipped another update to Stripe Projects: more agents, more providers, more developer controls. But the thing I keep coming back to isn't the feature list, it's who we're building with. Automattic. Wix. Laravel. ClickHouse. HeyGen. E2B. Each one a team that looked at the agentic era and said: the dev stack needs to be programmable, not just configurable. And they built toward that alongside us. That's the best part of this job. We're not just building Stripe features, we're building two-sided infrastructure. The network only works if both sides win: developers get a unified way to spin up their entire stack from a CLI, and providers get distribution to the engineers and agents who need them most. 2026 is the year agents become builders. We're trying to make sure the infrastructure is ready when they show up. https://lnkd.in/g6C5KDgu
95
8 Comments
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content