<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Scott Pallas on Medium]]></title>
        <description><![CDATA[Stories by Scott Pallas on Medium]]></description>
        <link>https://medium.com/@scottpallas?source=rss-bec50534e06d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*I8qimYjSuJbe_tkerx_Uhw.jpeg</url>
            <title>Stories by Scott Pallas on Medium</title>
            <link>https://medium.com/@scottpallas?source=rss-bec50534e06d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 22:56:12 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@scottpallas/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Ten Minutes of Sunshine]]></title>
            <link>https://scottpallas.medium.com/ten-minutes-of-sunshine-e5ec5407078c?source=rss-bec50534e06d------2</link>
            <guid isPermaLink="false">https://medium.com/p/e5ec5407078c</guid>
            <category><![CDATA[leadership]]></category>
            <dc:creator><![CDATA[Scott Pallas]]></dc:creator>
            <pubDate>Thu, 30 Apr 2026 02:22:03 GMT</pubDate>
            <atom:updated>2026-04-30T02:22:03.775Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-l2wgwp8-zmFFWe0IWq9Kw@2x.jpeg" /></figure><p>Haleakalā Maui — photo credit Scott Pallas</p><p><em>It was 77 degrees in Michigan on a Thursday afternoon in April. I had ten minutes between back-to-back meetings. I stepped out into the sunshine. And that’s when it hit me.</em></p><h3>The Grind</h3><p>This week was chaos. Endless meetings. Priority one issues. Planning for next year. All the emails coming in, all the needs of the people around you, the projects you’re trying to complete, the code you’re trying to get out the door. Early mornings. Late into the evening. You know the feeling.</p><p>And look, don’t get me wrong. It’s fun. We’re doing amazing stuff with amazing people. The work itself is genuinely good. That part hasn’t changed.</p><p>But it’s a grind. And it’s April in Michigan, which means we’re finally starting to move from winter to spring. We get maybe 50, 60 sunshine days a year if we’re lucky. So sunshine is a big deal around here.</p><p>I was at home today. Thursdays and Fridays we mostly work from home. Back-to-back meetings all day long. And I had maybe ten minutes between calls.</p><p>I stepped outside into the sunshine.</p><p>And that’s when it hit me.</p><h3>I’m going to go build this</h3><p>I have a principal engineer on my team who has never stopped growing. Every time something new comes along, he leans in. He learns it. And then he pushes on it until he finds something nobody else has found yet.</p><p>This week we had a one-on-one and he showed me what he’d been working on. We’ve been building out our <a href="https://scottpallas.medium.com/how-were-rebuilding-our-engineering-model-around-ai-10bf10a4dd4e">agentic context engineering practice</a>, stitching AI and context together through every step of the SDLC. From brainstorming to deployment.</p><p>What he’d been doing was taking it further. Leveraging Agent Browser from Vercel to validate PR output against the original development card. And on top of that, using the MCP server from our real user monitoring system (RUM) to research production friction points.</p><p>Incredible. But then we started riffing.</p><p>What if we pulled those friction points into GitHub as issues automatically? And then built another agent to trigger off the new issue, research the repo, find where the bug lives, and submit a potential fix for a senior engineer to review?</p><p>From there it kept going. What if we extended RUM into our lower environments, which we don’t do today? And then built agents to simulate actual traffic against it. Agents that surf our e-commerce site, place orders, look for friction points. All in the lower environment, before anything ever gets to production.</p><p>Lower environments never match production. Different traffic. Different volume. Every engineer reading this knows the feeling. And here we were, talking about a way to close that gap with agents.</p><p>He was already pulling up his notes before we’d even finished the thought. He looked up and said, “I’m going to go build this.”</p><p>And there it is. Jumping right in at the chance to take it to the next level. I can’t wait to see where it goes.</p><h3>The Two-Way Street</h3><p>Earlier in the week I got a message from someone who used to be on my team. They moved to another part of the organization a while back, always did great work, and we’ve kind of stayed in touch.</p><p>They wanted to pick my brain. Organizational design, how to think about structuring a growing team, how to manage inconsistent demand from across the business.</p><p>Honestly? I was honored to be asked.</p><p>We got together and I spent a lot of time just asking questions first. What’s coming in? How they’re organized? Are there themes? Is there long-running work versus ad hoc stuff? What does the funding look like?</p><p>Then we talked through it. My advice was to invest where you know the work is going to be consistent over time. Build dedicated teams around those areas. And for the inconsistent, ad hoc, hot-priority stuff, build a focused team that can flex to meet that demand without destabilizing the teams doing the steady work. When it comes to funding, set a base amount for the dedicated teams. Create a process for ad hoc projects to bring their own funding. That keeps skin in the game.</p><p>Did we also derail into a little bit of context engineering? Of course we did.</p><p>People sometimes think mentorship is a one-way street. Like the person asking is getting something and the person giving is just giving.</p><p>That’s not how it works. At least not for me.</p><p>You sit across from someone. You learn their world. You understand what they’re trying to build. That’s valuable. I walked away having learned something too.</p><p>Nobody actually knows exactly what to do in every situation. You draw on what you’ve lived. You talk. You listen. You learn from each other. That’s where useful stuff comes from.</p><p>For both people.</p><h3>Five Years. Seven Months.</h3><p>The third thing that hit me this week was a project update.</p><p>We are in the middle of rewriting our entire front end. A project that touches everything. What we originally built took five years. We’re rebuilding it in seven months with our <a href="https://scottpallas.medium.com/how-were-rebuilding-our-engineering-model-around-ai-10bf10a4dd4e">new way of working</a>.</p><p>Two months in, I started getting the updates that showed we’re actually doing it. Not in theory. Not on a slide. In the code.</p><p>Are we spiking the ball? No. There’s still a long way to go. Several waves of MVP ahead of us.</p><p>But two months in, the thing we believed was possible is starting to be real.</p><h3>Ten Minutes</h3><p>So I stepped outside today between meetings. 77 degrees. Sunshine. Michigan in April, which if you live here you know is basically a gift.</p><p>And I just stood there for a few minutes and let it land.</p><p>It’s the little things. The progress. The growth of the people around you. The conversations you didn’t see coming.</p><p>That’s what makes the grind worth it.</p><p>The moments you almost miss because you’re too heads-down to step outside.</p><p>Ten minutes of sunshine. That’s all it took.</p><p>Read the previous post in this <a href="https://scottpallas.medium.com/how-were-rebuilding-our-engineering-model-around-ai-10bf10a4dd4e">series</a>.</p><p><em>When not coding into the early morning hours, Scott leads digital e-commerce engineering at a large retailer and in the warmer months can occasionally be found spending time on the Great Lakes.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e5ec5407078c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How we're rebuilding our engineering model around AI]]></title>
            <link>https://scottpallas.medium.com/how-were-rebuilding-our-engineering-model-around-ai-10bf10a4dd4e?source=rss-bec50534e06d------2</link>
            <guid isPermaLink="false">https://medium.com/p/10bf10a4dd4e</guid>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[leadership]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[future-of-work]]></category>
            <dc:creator><![CDATA[Scott Pallas]]></dc:creator>
            <pubDate>Mon, 09 Mar 2026 16:31:00 GMT</pubDate>
            <atom:updated>2026-03-09T16:31:00.678Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VLfVgq5DWyf5ds5lPdxxIA.jpeg" /><figcaption>Lake Michigan — Photo Credit: Anya Pallas</figcaption></figure><h3>Today Is the Day That Starts It All</h3><p>Today is special.</p><p>Today is the day we re-energize the team. Today is the day we’re all going to look back on. The one where it started.</p><p>I don’t say that lightly. I’ve been in enough rooms where someone declared a fresh start and nothing actually changed. This is different. Today something real is shifting, and I want to talk about it while the ink is still wet.</p><p>A few weeks ago I wrote about how <a href="https://medium.com/@scottpallas/getting-back-into-code-ai-rekindled-my-love-for-programming-db62d2d45bb0">AI rekindled my love for coding</a>. That was personal. This one is professional. And the two are more connected than you’d think.</p><h3>We Don’t Have All the Answers. That’s the Point.</h3><p>I want to be upfront about something before I get into the model we’re trying.</p><p>We don’t know if this is the exact right way to go. We’re not claiming we’ve figured it out. What we do know is that this moment in time is a real opportunity, and if we don’t try something different, we’ll look back and wonder why we played it safe when the game was changing around us.</p><p>So we’re trying. We’re betting on our people and on this moment. And we’re going to learn out loud.</p><h3>The People Come First</h3><p>Before I talk about smaller teams and sprint cadence and context engineering, I want to say the thing that actually matters most to me.</p><p>We have amazing people. Across the board. Engineers, leaders, designers, devops, QA. People who care deeply about what they build and who they build it for. That hasn’t changed and it’s not going to change.</p><p>What today is really about is giving those people an unlock.</p><p>We’ve always given our team freedom with how they work, but the complexity of our systems and process made experimentation expensive. AI engineering tools change that. The cost of trying something is lower, the speed of recovery is faster, and there are fewer reasons to play it safe.</p><p>The message I want our team to hear today is: it’s okay.</p><p>It’s okay to experiment. It’s okay to try something and have it not work. It’s okay to move fast and learn fast and share what you learned so the person next to you doesn’t have to start from zero. Psychological safety isn’t a soft concept. It’s what makes high-performing teams actually perform. And we’ve got the kind of people who will absolutely run with it if we give them the space.</p><p>That’s what today is really about.</p><h3>The Model</h3><p>So here’s what we’re trying.</p><p>Smaller, highly specialized teams. Built around AI from the start, not retrofitted. A focused team of three to four with the right tooling, the right context, and a weekly sprint cadence can move in ways that larger teams simply can’t. I’ve watched it happen. We’re structuring to bet on it.</p><p>Weekly sprints aren’t just a process choice. They’re a forcing function. If you can’t close something meaningful in a week, one of two things is true: the work is too big, or the context isn’t clear enough. Either way, you find out fast. That’s the point.</p><h3>Context Engineering</h3><p>Here’s the thing I keep coming back to, and I don’t hear enough people talking about it.</p><p>AI tools are only as good as the context you give them.</p><p>We’re investing in context engineering. Making sure every story, every pull request, every system interaction has the right information for the AI tools working alongside the team. This isn’t just writing better prompts. It means acceptance criteria that leave nothing ambiguous. Stories that contain the architectural decisions, the edge cases, the why behind the what.</p><p>When a developer sits down with Claude Code and the context is right, they don’t spend 30 minutes re-explaining the codebase. They just move.</p><p>When we inspected every step of the process, it became clear how far upstream this thinking has to go. The person writing a well-contextualized story is doing as much to enable AI-powered engineering as the engineer using it. That’s not just an engineering mindset shift. It’s a shift for the whole team.</p><h3>Speed is a System Problem</h3><p>You can invest heavily in AI-powered engineering and still not move fast. If you speed up one part of the lifecycle and leave everything else the same, you’ve just moved the bottleneck. You haven’t removed it.</p><p>Think about what actually happens end-to-end.</p><p><strong>Stories.</strong> If a story takes two weeks to get to ready and engineering finishes it in three days, you haven’t won. You’ve just made the upstream problem impossible to ignore. We’re applying the same AI-first thinking to how stories get written and refined. Speed has to start before the first line of code.</p><p><strong>PR reviews.</strong> More code, faster engineering, means more PRs. If review cycles don’t keep up, the human layer becomes the new bottleneck. We’re investing in a small team focused specifically on moving code through the pipeline, leveraging AI and automation to handle the mechanical checks so that when a human reviewer shows up, they’re focused on judgment calls, not checklists.</p><p><strong>UX and design.</strong> Speed means nothing if the experience isn’t right. We have an amazing UX team, always focused on the customer experience. We’ve added a small UX enabler team focused specifically on giving our designers better tools and faster feedback loops, so their work can keep pace with the speed we’re building everywhere else.</p><p><strong>Testing and quality.</strong> Shift-left, automated test generation, AI-assisted triage. In progress. Because if QA is the last thing you speed up, it becomes the next constraint.</p><p>The goal isn’t faster engineering in isolation. It’s faster everything, moving together.</p><h3>Demo the Work. Demo the Process. Demo the Learnings.</h3><p>Here’s something we’re doing that I think matters as much as anything else.</p><p>We’re not just going to demo what we build. We’re going to demo how we built it. The process, the decisions, the things that didn’t work, the things that surprised us. We want every person on our team to get their hands on it.</p><p>One team. Moving forward together. Because if only a handful of people level up and the rest are watching from the sidelines, we haven’t actually transformed anything.</p><p>I don’t know where this goes. 10x? 100x? I don’t know. But I’m more excited about the direction we’re pointed than I’ve been in a long time, and I think our people are going to surprise us.</p><h3>Why Today</h3><p>I’ve been in tech leadership long enough to have a healthy skepticism of transformations. Most of them are process changes dressed up as something bigger.</p><p>This one feels different because the underlying capability has actually changed. Not a little. A lot.</p><p>In January, Cursor’s CEO revealed that a swarm of AI agents, running autonomously for one week, built a web browser from scratch. Over three million lines of code. A custom rendering engine. It kind of works. That’s not an edge case. That’s a preview.</p><p>The companies that reshape around this will move at a pace that’s hard to catch. The ones that bolt AI onto their existing process and call it a transformation will spend the next few years wondering why the gap keeps widening.</p><p>We’re choosing to reshape. Smaller teams, full AI integration, context engineering as a real discipline, and a relentless focus on the whole lifecycle.</p><p>Today we start. I don’t know exactly where it goes. But I know what today is.</p><p>Today is the day that starts it all.</p><p>Stay tuned. I’ll be writing about the wins, the losses, and the stuff nobody talks about.</p><p><a href="https://scottpallas.medium.com/getting-back-into-code-ai-rekindled-my-love-for-programming-db62d2d45bb0">Read the first post in this series</a></p><p><em>When not coding into the early morning hours, Scott leads digital e-commerce engineering at a large retailer and in the warmer months can occasionally be found spending time on the Great Lakes.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=10bf10a4dd4e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Getting back into code: AI rekindled my love for programming.]]></title>
            <link>https://scottpallas.medium.com/getting-back-into-code-ai-rekindled-my-love-for-programming-db62d2d45bb0?source=rss-bec50534e06d------2</link>
            <guid isPermaLink="false">https://medium.com/p/db62d2d45bb0</guid>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[coding]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[claude]]></category>
            <dc:creator><![CDATA[Scott Pallas]]></dc:creator>
            <pubDate>Sat, 21 Feb 2026 01:14:21 GMT</pubDate>
            <atom:updated>2026-02-21T01:14:21.987Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7ro5fiz5nr2k9TeobLNSAA.jpeg" /><figcaption>Utah Skiing — Photo by Brooklyn Pallas</figcaption></figure><p><em>A few weekends ago I was on vacation with my family in Utah. We were skiing, hiking and enjoying some needed time away. It’s 2am on Saturday night and I’m grinning from ear to ear. I had just completed a fully functioning e-commerce site deployed to production. Eighteen AI agents built while I directed traffic. This is what coding feels like now.</em></p><h3><strong>Back In The Day</strong></h3><p>My love for programming began at a young age. I wrote my first lines of code when I was twelve, using the IBM pc my father brought home from work. You could say I was not the kid that applied myself in school and the result was a bit of being grounded from just about everything. But here’s the thing, my father never grounded me from using the computer. By eighth grade I had taught myself some programming. These were some good times using basic, pascal and assembler languages. Coding was my happy place and he let me keep it.</p><p>From there it was on to running bulletin board systems and traversing compuserve and other early online systems (does anyone remember these?).</p><p>Fast forward into my early career years I was coding in c/ c++. Dealing with pointers, memory management and debugging late nights and into the early morning. I spent years building things the hard way because there was no other way but I loved it. There’s something super satisfying about software creation.</p><p>Then life happened. My career moved into technology leadership. I’ve spent the last couple decades leading engineering teams, setting the strategy and bringing amazing solutions to life. And I am still having the time of my life doing it. I truly wake up every day looking forward to what I do. The one gap is the distance between me and the code has grown wider every day. Until now.</p><h3><strong>The Spark</strong></h3><p>Two of my colleagues changed everything! About 6 months ago they introduced me to Claude code. I had already been all in on using AI in my leadership role. Spending my days in the office using Microsoft co-pilot for just about everything by day and using ChatGPT and Perplexity at home. That introduction to Claude code, using the CLI, and various AI powered tools triggered something deep inside me. Between that and the passion they shared, I jumped in. There is no turning back. Those two lit the spark and have changed my life. (You know who you are — thank you!)</p><h3><strong>Coding is fun again</strong></h3><p>AI coding tools are moving at a breakneck pace. It’s only been about a year since Claude Code launched and the continued advancements are staggering. I haven’t coded like this since my twenties. Every evening after the day job wraps, I’m back at it. Building real projects. Staying up way too late because I’m simply having too much fun to stop. My wife is thrilled, as you can imagine.</p><p>You know that list everyone has? The one with the various ideas scribbled on paper or in various online note taking systems? Ideas you’ve had, things you’ve thought of, never took action on and missed? I have a long list. AI has changed the landscape completely. I’m finally tackling some of those ideas one by one.</p><p>Here’s what is different now versus twenty years ago: the ratio of ideation and software creation versus fighting with syntax and tooling has completely flipped. In my c / c++ days I’d spend more time debugging, finding information and trying to remember the proper syntax. Now I spend most of my time on architecture, the user experience, and actually bringing these ideas to life. The AI handles the tedious stuff and I get to focus on the parts that matter. And my documentation now is top notch (thanks Claude).</p><p>That doesn’t mean the AI does everything for you. I mean it can, but for the best results you still need to approach your project with a software engineer’s mindset. You still need the fundamentals. How systems connect, how data flows, how to think about performance, scalability and security. A few decades of technology leadership are not wasted here. If anything, that experience is more valuable than ever. I know what good software solutions look like at scale and if you’re like me you probably do too. The AI just helps me get there faster.</p><h3><strong>What Caught Me Off Guard</strong></h3><p>Coming back after this long, a few things surprise me.</p><p>The stack is actually simpler now. I expected modern development to be more complicated. In a lot of ways it’s not. Typescript gives you type safety without the pain of header files. Next.js handles routing, rendering and API endpoints in one framework. MongoDB lets you store data without writing SQL migrations. Platforms like Vercel make it easy to deploy your apps with a simple git push or better yet link to your repo. And several of these platforms have free tiers so it’s easy to jump back in.</p><p>Claude Code is a better pair programmer than most humans. That’s not a knock on my programmer friends. Claude is infinitely patient, available at all hours, never judges me for asking basic questions, and can jump from brainstorming to coding to debugging without missing a beat. It’s like having a senior engineer next to you that has access to all the documentation ever written. And never needs any sleep. Apparently neither do I.</p><p>The fundamentals still matter. Knowing c / c++ didn’t teach me React, but it taught me how to think about programs. Memory management taught me how to think about resources. Networking taught me how systems work together.</p><p>Deploying working code is addictive. When you can go from idea to deployed app in an evening, the dopamine loop is real. I’ve built more functional software in the last few months than I built in years of side project before. The speed creates momentum, the momentum creates more ideas. The notebook of ideas continues.</p><h3>Eighteen Agents and a Ski Lodge</h3><p>Bringing it back to that Saturday in Utah. I’d been experimenting with <a href="https://medium.com/@steve-yegge/welcome-to-gas-town-4f25ee16dd04">Gas Town</a>, <a href="https://medium.com/@steve-yegge">Steve Yegge’s</a> multi-agent orchestrator for Claude Code. If you haven’t heard of it, the short version is instead of one AI coding agent, you’re running a coordinated team of them. A mayor agent delegates the work to specialized workers, each handling different parts of the build. It’s like being a general contractor who just describes what the building should look like and the crew of fast workers go to work.</p><p>That’s a whole other blog post for a later date and I’ll have more stories to share. But the short version is I was lying in bed after a day in Utah, described what I wanted and watched 18 agents build it. Two hours and ten minutes. A real, working, deployed e-commerce site. Not a demo. A site you could buy things on.</p><p>Twenty years ago that would have taken a couple months with a full team. Now it’s a Saturday evening side quest while the family sleeps.</p><h3>Jump Back In</h3><p>If you’re like me, someone who used to code, drifted away, and has been watching from the sidelines, just jump back in. If you’re not like me but have always wanted to code, now is the time, jump in with me.</p><p>AI hasn’t replaced the need for developers. It’s removing the friction that kept people from building. The cognitive load of remembering all the syntax, the APIs, the configs. Now you have your pair programmer at your side to assist with all those details. That’s all handled now. What’s left is the good stuff: designing systems, solving problems and creating things people will use. And of course grinning from ear to ear.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*L60LKYJPAuKTPfHoDaxleA.jpeg" /><figcaption>Lake Michigan — Photo by Scott Pallas</figcaption></figure><p><em>When not coding into the early morning hours, Scott leads digital e-commerce engineering at a large retailer and in the warmer months can occasionally be found spending time on the Great Lakes.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=db62d2d45bb0" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>