James Yun

Ordered a hot dog. They said they didn’t have hot dog buns and offered to stuff the sausage into a burger bun. Sure. Who cares… This reminded me of the only time I’d been at a Michelin restaurant. Spent a lot, ate a little. It was okay, not worth the money. As Nassim Nicholas Taleb put it, fancy places have negative utility. There’s nothing better than a slice of pizza from a corner shop after a dinner at a fancy restaurant with small dishes and excessive rituals. In Russian, we have a saying “don’t create a cult out of food” which boils down to “eat simple”. Maybe that is why Russian cuisine is so bland and unpretentious. And I like it that way. So sausage on a burger bun, fries and 20% tip it is.

Last week, I had an incredible opportunity to travel to the International Center of Tropical Agriculture (CIAT) in Palmira, Colombia to deliver two technical workshops on Amazon SageMaker AI and HPC on AWS in support of CGIAR’s Enterprise Breeding System (EBS) Hackathon. This was a powerful engagement for me as I connected with staff from various organizations around the world, and gained valuable insights into how agriculture can intermingle with technology. Not to mention, this was my first time traveling out of the US! I’d like to thank Robert M Bata for coordinating things on our end, Andre Embersics and Peter Selby for coordinating the event, and Rubem Saldanha for partnering with me, delivering two additional sessions on AWS Infrastructure & research applications, and for giving me his perspective on Amazon’s peculiar culture. I am very grateful for the discussions I had with all participants of the event, and for this opportunity to support AWS business operations in another region!

Excited to share that I’ve graduated from the University of Virginia Darden School of Business. Darden’s Executive MBA program was demanding in the best possible way. The faculty brought a level of depth and dedication that genuinely raised the bar, matched only by the breadth of experience and perspective added to every discussion by an equally impressive student cohort. Two years, countless lessons, all worth it. Furthermore, words cannot express how truly thankful I am for the unsung heroes behind the scenes, both at home and in the office. It was their support, encouragement, and flexibility that made this opportunity possible. Sincerely grateful for the experience and looking forward to the next chapter!

The candidate was perfect on paper. Prepared, articulate, confident. I still said no. Here's why. Over the years of interviewing experience, I've seen a pattern that holds most candidates back. It's not the technical skills. It's not the preparation. They tell me WHAT they did. Not WHY they decided to do it. Simon Sinek nailed it in Start With Why. Most people communicate from the outside in. WHAT → HOW → WHY The most inspiring leaders and the most memorable candidates do the opposite. WHY → HOW → WHAT And honestly? The same holds true in every interview — not just Amazon. I ask: "Tell me about a time you took ownership of a problem." Most candidates walk me through a detailed timeline of events. The bug. The fix. The deployment. The result. Clean. Structured. Forgettable. What I'm actually listening for? → WHY did YOU step up when others didn't? → What trade-offs did you consciously make? → What did you learn that changed how you work? Amazon Leadership Principles aren't a checklist. They're a lens into your WHY. A strong STAR answer doesn't just tell me what happened. It starts from the inside out — purpose first, action second. The candidates who stand out? They don't just describe the situation. They make me feel the weight of the decision. If you're preparing for an interview (Amazon or any other company) - Stop rehearsing your stories. Start understanding WHY those stories reveal who you are. That's the difference between a good answer and a Bar-Raising one. "People don't buy what you do. They buy why you do it." — Simon Sinek https://lnkd.in/dSwJaRjV Have you read Start With Why? How has it changed how you communicate? Drop in the comment below #Amazon #Interviews #BarRaiser #LeadershipPrinciples #StartWithWhy #SimonSinek #GoldenCircle #CareerGrowth #Hiring #InterviewTips Disclaimer: Personal reflections as an engineering manager and Amazon Bar Raiser. Any details referred are based on publicly available information. #️⃣ Post 3

🚨 Most candidates never know this about Amazon interviews. One person in that room has a completely different job than everyone else. They're not there to hire you. They're there to protect the standard. That person is the Bar Raiser. 🔍 What Bar Raisers actually do? They're independent. Not from your hiring team. They bring an objective lens. Even if the entire panel says yes, a Bar Raiser can say no. 🛑 They're not looking for perfect answers. They're looking for signal: 🎯 Do you own your outcomes - good and bad? 🧠 Can you think structurally under pressure? 💡 Do you live the Leadership Principles or just recite them? 📈 Will you raise the bar, or just meet it? June 2025 — that's when I officially became a Bar Raiser. Since then, I've been in loops across teams, functions, and levels I'd never have touched otherwise. 200+ interviews over 5 years will do something to how you see people. You stop evaluating answers. You start reading patterns. 👀 The best candidates make you think. The rest make you wait. 📖 Stories from my Bar Raising practice: Story 1: When the room was already convinced 🤝 A candidate walked in with an impressive track record. The panel was sold. Leadership signals? Strong. Project management? Excellent. But I noticed something in the technical assessment that didn't add up. 🔎 Through focused follow-up questions, I helped the panel see a critical disconnect: Strong leadership credentials. Weak hands-on technical depth — for a role that demanded it. I redirected the conversation: What does this role actually require? The panel reached a consensus "not inclined" — going against the majority's initial read. ✅ Lesson: Past success is not a proxy for current role fit. Story 2: When "no" wasn't the right answer either 🔄 A candidate showed real strengths — global process implementation, cross-cultural stakeholder management, strong initiative. But those strengths didn't map to the strategic depth the role demanded. Instead of closing with a "not inclined," I asked: Where does this candidate actually belong? We recommended them for a different role — a better fit for their proven expertise. We maintained the bar. And we didn't lose strong talent. 🏆 ✅ Lesson: Raising the bar isn't just about saying no. It's about placing people where they'll truly excel. 💬 What bar raising really is? It's not gatekeeping. It's not being the toughest person in the room. It's asking the question no one else is asking: Will this person make us better? 🚀 **Clarity. Ownership. Depth** That's what separates great candidates from good ones. See you in the next post. 👋 https://lnkd.in/gQJtkKqu Disclaimer: Personal reflections as an engineering manager and Amazon Bar Raiser. Any details referred are based on publicly available information. #️⃣ Post 2

Today, Atalanta is coming out of stealth. We are a mathematical AI company with a vision to bring provably correct decision-making to the world's most important missions. The systems underpinning critical infrastructure, national security, and autonomous operations are growing more complex and interdependent than our ability to reason about them. The U.S. government calls this challenge "the software understanding gap." We built Atalanta to deliver on this national priority by combining AI and formal methods with the speed, agility, and precision that critical missions demand. Over the past 15 months, we've deployed our flagship product, Argo, with leading institutions across energy, space, and defense. We've built a team of exceptional people deeply committed to solving this problem, and we're now growing quickly and expanding across engineering, mathematics, strategy, and design. Read our company thesis in the essay "The Age of Software Understanding," written by my co-founder Jonathan Ring and me: https://lnkd.in/e2piRfWx

"Be pessimistic. Question everything." What 12 months of delays taught me.. 📌 PART III: FINAL MILE FAILURES (MISTAKES 9–11) 🔌 MISTAKE #9: WE UNDERESTIMATED INTEGRATION COMPLEXITY 200+ service integrations across 27 regions required far more coordination than anticipated. Many issues only appeared under real production load. ⚠️ Impact: Version incompatibilities required last-minute architectural changes. ✅ Lesson: Integrations are partnerships, not API calls. Start discussions 6+ months before launch. Design for partial failures. 📅 MISTAKE #10: WE IGNORED CALENDAR REALITIES We planned a major release during end-of-year without accounting for conferences and holiday vacations. ⚠️ Impact: Key team members unavailable during critical decisions. External dependencies nearly impossible to resolve. ✅ Lesson: Build a people-availability calendar alongside your technical roadmap. Avoid end-of-year and conference season releases. 🎭 MISTAKE #11: WE DIDN'T REHEARSE LAUNCH DAY We assumed the launch procedure was well-understood — we had never rehearsed it step by step. ⚠️ Impact: Configuration gaps revealed themselves on launch day at the worst possible moment. ✅ Lesson: Rehearse your launch day. The procedure is part of the product. Release critical config changes early. ⭐ WHAT I'D DO DIFFERENTLY → Launch smaller first, validate then expand. → Decouple launches. Establish API contracts and run joint integration testing. → Defend code freeze dates. 3–4 weeks of bake time is non-negotiable. → Avoid end-of-year releases. Treat availability calendars equal to technical milestones. → Be relentlessly pessimistic. Question every estimate and dependency. 3–4 months of planning saves 12+ months of delays. 🧠 THE MINDSET SHIFT Before: I optimized for speed — move fast, ship faster. After: I optimize for informed pessimism — because the cost of being wrong is always higher than the cost of being careful. Questions I now ask: "What are we not testing and why?" "Are we ready or do we just want to be?" "If this person left tomorrow, would we be okay?" The engineer who asks “what if this fails?” is not being negative. They’re being realistic. The leader who adds buffers is not being slow. They’re being responsible. "In distributed systems, shortcuts don't save time — they create technical debt that compounds with interest." 🎯 The Bottom Line Our launch succeeded — 99.99% availability, positive user sentiment. But delayed timelines and team stress were a price worth avoiding. The real lessons for me? Start smaller. Add more time. Question every assumption and dependency. Defend your process when pressure peaks. Pessimism isn't negativity, it's how resilient systems get built. Thank you for following along on this journey. Your engagement and shared experiences make these lessons even more meaningful. 🙏 Disclaimer: Personal reflections as an engineering manager. Technical details based on publicly available information. #️⃣ Post 1 - Part 3