Three principles for sustainable AI adoption: a guide for engineering leaders
A guide to bringing agentic LLM tools like Claude Code into an engineering organization, focusing on supporting your team and investing in engineering fundamentals.
Recent content on Thoughts on Systems
A guide to bringing agentic LLM tools like Claude Code into an engineering organization, focusing on supporting your team and investing in engineering fundamentals.
Nono is a new lightweight sandboxing tool that takes advantages of native Linux and macOS kernel-level isolation controls to wrap Claude Code and other agentic coding harnesses (or any process) and prevent them from deleting your home directory or dropping your database.
OpenAI engineer Ryan Lopopolo writes in Harness engineering: leveraging Codex in an agent-first world on how to effectively use coding tools by managing context.
Thariq Shihpar, an engineer at Anthropic, breaks down how Prompt caching is everything in building LLM apps such as Claude Code.
Chris Lattner, who created projects like LLVM that back modern open-source compilers, reflects on What the Claude C Compiler Reveals About AI Coding .
Michael Bolin provides a detailed explainer Unrolling the Codex agent loop | OpenAI on the OpenAI blog.
Zoë Hitzig writes that OpenAI Is Making the Mistakes Facebook Made. I Quit.
Cedric Chin recently started a series on The Study of Asian Tycoons . How are Asian conglomerates different from Western businesses?
Harper Reed documents his current LLM codegen workflow : effectively using LLMs in an agentic way via different prompts. He first uses an architect and then a project planner and finally a software engineer.
Charity Majors takes on the recent rollbacks of DEI policies at major companies, writing that Corporate “DEI” is an imperfect vehicle for deeply meaningful ideals .
Declan Chidlow expresses concern that AI is stifling tech adoption .
How did DeepSeek achieve such remarkable capabilites in their recent models? In various interviews, founder Liang Wenfeng explains his approach on staffing a culture of innovation.
Jon Keegan notes in Sherwood News that A free, powerful Chinese AI model just dropped — but don’t ask it about Tiananmen Square , exploring what chain-of-thought models reveal about the perspectives encoded in their training.
Molly White educates us again on how to think about cryptocurrency valuations, writing in her latest Citation Needed newsletter that, No, Trump didn’t make $50 billion from his memecoin
Jack Danger explains the concepts of Infrastructure Gravity & Domain Engineering in an excerpt from his book, Executive Engineering .
Lena Cohen writes for the Electronic Frontier Foundation that Online Behavioral Ads Fuel the Surveillance Industry — Here’s How . Joseph Cox of 404 Media dives deeper on Wired.com to identify specific apps that contribute to surveillance: Candy Crush, Tinder, MyFitnessPal: See the Thousands of Apps Hijacked to Spy on Your Location .
Derek Thompson writes in The Atlantic about The Anti-Social Century .
Benji Webber provides a list of tips for overcoming resistance to Extreme Programming .
Simon Willison summarizes key learnings about LLMs in 2024 .
Startups engineers often make a false dichotomy between moving fast and writing clean code, taking on unnecessary tech debt. Improving skills and good leadership can manage tech debt and keep code maintainable.
Startups can often find themselves firefighting due to technical debt. Here are 4 steps to start pulling your company to a more proactive state.
Too often quality focuses on reducing bugs or crashes. This leads to a focus on avoiding errors. Instead, seek to create a culture and organization that is focused on achieving excellence.
Quality in products requires more than test coverage; engineers benefit from a deep understanding of the problem domain and user personas.
A good resume can help you get noticed by a hiring manager. Here are a few tips that might help you improve your resume.
Our productivity is often limited by how quickly and effectively we can get better at figuring out what’s going on and moving to a solution. Some suggestions on things to study and how to study them.
Sooner or later, every technology organization faces a necessary evil—migrations. Make yours go smoothly by understanding the old system well, engaging with all of your stakeholders early, tracking behavioral differences, and being ready to iterate.
You can’t learn everything on the job. Lead for the team. Overcommunicate. Lean on people smarter than you.
How do we train professional musicians compared to software engineers? Musicians develop an ability to relate theory and history to practice; we should do more in computer science.
Some thoughts on how to ensure systems research is reproducible by sharing code.
The Chord protocol dynamically constructs robust and scalable overlay networks that map a given key to an active node. Here are some learnings from implementing and operating Chord networks in the wide area.
When a graduate student completes their PhD, the software they wrote for that degree begins an almost inexorable decline into obscurity. Other things become more important and, unless that code serves as a platform for further research or has spun off into a startup, there is precious little time to maintain and further develop code written for the degree. The code has already served a…
Many corporate firewalls prevent git from using its efficient binary protocol by blocking outbound network connections. Sometimes, you are lucky and are trying to clone a repository that is hosted on a site like github which exports their repositories over HTTP , which would enable you to get through the firewall using the http_proxy environment variable. However, you are usually not that lucky…
Tim Daly , one of the developers of Axiom , has a vision for solving the following problem: Computational science seems to be in a state where people work independently. They develop whole systems from scratch just to support their research work. Many make an effort to distribute their system but fail for lack of interest. Worse yet, the research that gets published often cannot be used by others…
In the course of my research, I tend to do a fair amount of data analysis and reduction. This ranges from simple statistics to in-depth examination of traces . While working on the camera-ready for our NSDI paper, I found myself thinking about Vern Paxson ’s IMC paper on Strategies for Sound Internet Measurement (PDF . The paper is filled with good material and I long for the kinds of…
Work # I lead and advise engineering organizations. My working style prioritizes developing leadership at all levels to facilitate aligned but decentralized execution. Strong engineering fundamentals and people development are what enable teams to deliver high-quality products. I helped shape the growth of the engineering organization at Toast from ~30 engineers in 2016 to nearly 1000 in 2024. The…
Emil Sit offers consulting on leadership and management