Inference Engineering with Baseten's Philip Kiely

In this episode, Scott talks with Don Syme about the emerging world of agentic developer workflows and what it means when coding tools move from autocomplete helpers to collaborators. They explore how modern tools like GitHub Copilot and GitHub Agentic Workflows are evolving into systems that can plan, execute, and iterate on tasks across a codebase, and what that means for software design, type systems, and developer responsibility.  https://github.github.com/gh-aw/

This week on the show, Scott talks to Philip Kiley about his new book, Inference Engineering. Inference Engineering is your guide to becoming an expert in inference. It contains everything that Philip has learned in four years of working at Baseten. This book is based on the hundreds of thousands of words of documentation, blogs, and talks he's written on inference; interviews with dozens of experts from our engineering team; and countless conversations with customers and builders around the world. https://www.baseten.co/inference-engineering/

What does it take to design a programming language from scratch when the target isn’t just CPUs, but GPUs, accelerators, and the entire AI stack? In this episode, I sit down with legendary language architect Chris Lattner to talk about Mojo — his ambitious attempt to rethink systems programming for the machine learning era. We trace the arc from LLVM and Clang to Swift and now Mojo, unpacking the lessons Chris has carried forward into this new language. Mojo aims to combine Python’s ergonomics with C-level performance, but the real story is deeper: memory ownership, heterogeneous compute, compile-time metaprogramming, and giving developers precise control over how AI workloads hit silicon. Chris shares the motivation behind Modular, why today’s AI infrastructure demands new abstractions, and how Mojo fits into a rapidly evolving ecosystem of ML frameworks and hardware backends. We also dig into developer experience, safety vs performance tradeoffs, and what it means to build a language that spans research notebooks all the way down to kernel-level execution.

There’s a new wave of AI tools that don’t just live in the cloud, don’t just autocomplete code, and don’t just sit in a browser tab. They reach into your local environment, understand your context, and act more like a thinking companion than a chatbot. In this episode, I talk with Peter Steinberger, founder of OpenClaw, about the rise of “The Claw” and what it means to build AI that feels fast, personal, and deeply integrated into your workflow. We explore why OpenClaw is having a moment, how developer expectations are shifting from prompts to agents, and what it takes to design tools that balance power, safety, and usability. Peter shares the architectural choices behind OpenClaw, the tradeoffs between local and cloud inference, and his perspective on privacy, ownership, and latency in a world of ever-larger models. This is a conversation about control. Who owns your context? Where does your data live? And what happens when AI stops being a destination and starts becoming an ambient layer across everything you do?

AI is making developers dramatically more productive...so why is everyone so exhausted? In this episode, Scott talks with Steve Yegge, legendary blogger and creator of Gas Town, a multi-agent orchestrator he describes as "Kubernetes for coding agents." Steve shares his theory of the "AI Vampire," that working alongside AI drains human energy Colin Robinson-style (What We Do In The Shadows), even as output skyrockets. They dig into what happens when you're managing ten or twenty Claude Code instances at once, who actually captures the value of a 10x productivity boost, and why the most important thing developers can do right now might be to close the laptop and go for a walk.

Code reviews are one of the most powerful tools teams have for maintaining quality — but they're also one of the most emotionally charged parts of the development process. With AI coding agents generating more code than ever, the review bottleneck is growing fast. But what if AI-assisted reviews could not only keep up with the volume, but actually be kinder about it? Scott talks with Nnenna Ndukwe, Developer Relations Lead at Qodo, about how AI code review is evolving beyond glorified linting into something that understands context, catches what matters, and delivers feedback developers actually want to read. They explore what happens when the same AI writes and reviews its own code, and whether thoughtful AI review can make code review culture healthier for everyone...not just faster.https://www.qodo.ai/

Sandboxing is having a moment. As agents move from chat windows into terminals, repos, and production-adjacent workflows, the question is no longer “What can AI generate?” but “Where can it safely run?” In this episode, Scott talks with Mark Cavage, President of Docker, about the resurgence of sandboxes as critical infrastructure for the agent era and the thinking behind Docker’s newly released sandbox feature.They explore why isolation, reproducibility, and least-privilege execution are becoming table stakes for AI-assisted development. From protecting local machines to enabling trustworthy automation loops, Scott and Mark dig into how modern sandboxes differ from traditional containers, what developers should expect from secure agent runtimes, and why the future of “AI that does things” will depend as much on boundaries as it does on model capability.Learn more about Docker Sandboxes here: https://dockr.ly/4avCKTW

AI is moving faster than our collective ability to metabolize it. Between copilots, agents, vibe coding, and the ever-shifting definition of “senior engineer,” developers are asking a deeper question. Where is this all actually going? In this episode, Scott sits down with Gergely Orosz, author of The Pragmatic Engineer and longtime observer of how software gets built inside high-performing teams, to separate signal from hype.They dig into what AI is really doing to day-to-day engineering work. Productivity boosts versus skill atrophy. The changing expectations for junior developers. Whether “AI-first” companies are structurally different or simply marketing-forward. Gergely brings his trademark data-driven pragmatism, grounded in conversations with hundreds of engineering leaders navigating hiring freezes, agent experiments, and the reshaping of career ladders.Scott and Gergely also explore the human side. What happens to craftsmanship when code is abundant. How we teach the next generation to think, not just prompt. Why developer experience may matter more, not less, in an AI-accelerated world. Along the way, they consider whether we are watching a platform shift on the scale of cloud and mobile, or something even bigger.https://www.pragmaticengineer.com/

Join Scott and Eric Lippert for a lively tour through Fabulous Adventures in Data Structures and Algorithms, a fresh take on timeless topics that flips the script on how programmers think about core tools of the trade. Eric shares why he wrote a book that avoids the predictable interview-prep regurgitations, and instead dives into clever, lesser-known data structures and algorithmic ideas that he’s encountered over a long career in language design and tooling. You’ll hear how immutability can make data structures both simpler and faster, why backtracking shows up everywhere from tree search to puzzle solving, and how a deeper understanding of performance and abstraction can change the way you architect code. Along the way Eric reveals how to reconnect joy with problem solving, find surprising patterns that scale across domains, and build intuition that serves you long after the syntax fades from memory. https://www.manning.com/books/fabulous-adventures-in-data-structures-and-algorithms

Modern computers are faster than ever, yet much of our software feels slower, heavier, and more frustrating to use. In this episode of Hanselminutes, Scott talks with Vjekoslav Krajačić, creator of File Pilot, about bringing speed and responsiveness back to everyday tools.Vjekoslav built File Pilot as a reaction to bloated file managers and laggy interfaces, focusing on instant feedback, keyboard-first workflows, and a UI that feels immediate. We talk about what actually makes software feel fast, why modern frameworks often work against that goal, and how users instinctively know when an app respects their time.This is a conversation about restraint, craft, and why fast UIs still matter.https://filepilot.tech