Pretty much every company I know is looking for a way to benefit from Large Language Models. Even if their executives don’t see much applicability, their investors likely do, so they’re staring at the blank page nervously trying to come up with an idea. It’s straightforward to make an argument for LLMs improving internal efficiency somehow, but it’s much harder to describe a believable way that LLMs will make your product more useful to your customers. I’ve been working fairly directly on meaningful applicability of LLMs to existing products for the last year, and wanted to type up some semi-d ..read more

One of the most interesting questions I got after joining Calm in 2020 was whether Calm was a technology company. Most interestingly, this question wasn’t coming from friends or random strangers on the internet, it was coming from the engineers working there! In an attempt to answer those questions, I wrote up some notes, which summarize two perspectives on “being a technology company.” The first perspective is Ben Thompson’s “Software has zero marginal costs.” You’re a technology company if adding your next user doesn’t create more costs to support that user. Yes, it’s not really zero, e.g. S ..read more

At a recent offsite with Carta’s Navigators, we landed on an interesting topic: leadership roles sometimes mean that making progress on a professional initiative requires taking some personal risk. This lesson was hammered into me a decade ago during my time at Uber, where I kicked off the Uber SRE group and architectured Uber’s self-service service provisioning strategy that defined Uber’s approach to software development (which spawned a thousand thought pieces, not all complimentary). I did both without top-down approval, and made damn sure they worked out. It wasn’t that I was an anarchist ..read more

When you’re driving a car down a road, you might get a bit stuffy and decide to roll your windows down. The air will flow in, the wind will get louder, and the sensation of moving will intensify. Your engine will start working a bit harder–and louder–to maintain the same speed. Every sensation will tell you that you’re moving faster, but lowering the window has increased your car’s air resistance, and you’re actually going slower. Or at minimum you’re using more fuel to maintain the same speed. There’s nothing that you didn’t already know in the first paragraph, but it remains the most common ..read more

I recently got an email asking about self-publishing books, and wanted to summarize my thinking there. Recapping my relevant experience, I’ve written three books: An Elegant Puzzle was published in 2019 as a manuscript by Stripe Press (e.g. I wrote it and then it was released as is), which has sold about 100,000 copies (96k through the end of 2023, and selling about 4k copies a quarter over past two years), Staff Engineer which I self-published in 2021, which has sold about 70,000 copies (also selling roughly 4k copies a quarter over the past two years) The Engineering Executive’s Primer whic ..read more

Quick update on The Engineering Executive’s Primer. The book went to print yesterday, and physical copies will be available in March. Also, as of this moment, you can purchase the digital edition on Amazon, and read the full digital release on O’Reilly. (You can preorder physical copies on Amazon as well ..read more

Recently, I’ve thinking about where I want to focus my angel investing in 2024, and decided to document my thinking about value accumulation in artificial intelligence because it explains the shape of my interest–or lack thereof–in investing in artificial intelligence tooling. I’ll describe my understanding of the current state, how I think it’ll evolve over the next 1-3 years, and then end with how that shapes what I’m investing in. My view on the the state of play today: There are three fundamental components: Infrastructure (cloud providers, NVIDIA, etc), Modeling & Core (OpenAI, Anthr ..read more

In some pockets of the industry, an axiom of software development is that deploying software quickly is at odds with thoroughly testing that software. One reason that teams believe this is because a fully automated deployment process implies that there’s no opportunity for manual quality assurance. In other pockets of the industry, the axiom is quite different: you can get both fast deployment and manual quality assurance by using feature flags to decouple deployment (shipping the code) and release (enabling new functionality). The deeper I get into my career, the more I believe that example h ..read more

Perceiving the layers of context in problems will unlock another stage of career progression as a Staff-plus engineer, but there’s at least one essential skill to develop afterwards: navigating ambiguity. In my experience, navigating deeply ambiguous problems is the rarest skill in engineers, and doing it well is a rarity. It’s sufficiently rare that many executives can’t do it well either, although I do believe that all long-term successful executives find at least one toolkit for these kinds of problems. Before going further, let’s get a bit less abstract by identifying a few examples of the ..read more

One of the recurring debates about senior engineering leadership roles is whether Chief Technology Officers should actively write code. There are a lot of strongly held positions, from “Real CTOs code.” at one end of the spectrum, to “Low ego managers know they contribute more by focusing on leadership work rather than coding.” There are, of course, adherents at every point between those two extremes. It’s hard to take these arguments too seriously, because these values correlate so strongly with holders’ identities: folks who believe they are strong programmers argue that CTOs must code, and ..read more