Using LLMs effectively isn't about prompting
February 21, 2025 │ It's about building a sense of what LLMs do well
To avoid being replaced by LLMs, do what they can't
February 15, 2025 │ What LLMs can't do yet
Learning how to think better from LLMs
February 15, 2025 │ How LLMs approach problems, and what we can learn from it
Engineers who won’t commit force bad decisions
February 10, 2025 │ Why remaining non-commital is cowardly
Good engineers are right, a lot
February 6, 2025 │ It's an iffy Amazon leadership principle but a great engineering metric
How I use LLMs as a staff engineer
February 4, 2025 │ What I use them for and what I don't
Why does AI slop feel so bad to read?
February 1, 2025 │ A general theory of human/human and human/LLM interaction
Are DeepSeek's new models really that fast and cheap?
January 31, 2025 │ People are too confident that DeepSeek V3/R1 are 10x cheaper than 4o/o1
Why AI labs offer so many different models
January 28, 2025 │ A taxonomy of base models, distilled models, and reasoning models
What did DeepSeek figure out about reasoning with DeepSeek-R1?
January 26, 2025 │ RL-based reasoning instead of SFT on millions of traces
Playing politics is how senior engineers protect their team
January 26, 2025 │ If you're senior or above, you're morally obliged to play the game
Working fast and slow
January 25, 2025 │ How I deal with being way more productive on some days than others
Why are big tech companies so slow?
January 24, 2025 │ It's not incompetence or process, it's thousands of feature interactions
Protecting your time from predators in large tech companies
January 19, 2025 │ Some people will absorb as much work as you're willing to give them
Thinking clearly about software
January 18, 2025 │ To think clearly, think in slow motion
Giving pushback to leadership
January 16, 2025 │ Always give options, even when there aren't any
Crushing JIRA tickets is a party trick, not a path to impact
January 14, 2025 │ Tech companies care about projects, not tickets
Writing a tech blog people want to read
January 10, 2025 │ What I think about when I write blog posts
What it's like working for American companies as an Australian
January 10, 2025 │ My experience working across timezones for Americans
What makes strong engineers strong?
January 9, 2025 │ Self-belief, pragmatism, speed, and technical ability
Is it cynical to do what your manager wants?
January 8, 2025 │ Cynicism and optimism in software companies
Ratchet effects determine engineer reputation at large companies
January 5, 2025 │ Why you can't skip to the top (but you can skip to the bottom)
Mistakes engineers make in large established codebases
January 2, 2025 │ What I've learned from ten years working on projects like this
Glue work considered harmful
December 16, 2024 │ Why companies don't reward glue work and why you shouldn't do it all the time
How I got promoted to staff engineer twice
December 28, 2024 │ What does it take to get promoted from senior to staff at a tech company?
What can strong engineers do that weak engineers can't?
December 27, 2024 │ Engineering talent is about task-capability, not speed or volume
Why some engineers get trusted with high-impact work
December 22, 2024 │ Thoughts on building and maintaining trust with very senior managers
Grifters, believers, grinders, and coasters
December 2, 2024 │ Four archetypes in software engineering
I don't know how to build software and you don't either
November 20, 2024 │ Overconfidence about the big questions in software engineering
How I ship projects at big tech companies
November 9, 2024 │ What I think about when I'm lead engineer on a project
Talking to ChatGPT costs 5ml of water, not 500ml
October 28, 2024 │ Summarizing a whole lot of internet argument about the water usage impact of language models
MCTS and LLMs: what's the big deal?
June 17, 2024 │ Why people are excited about Monte Carlo tree search and language models
From hours to seconds: AI tools to detect animal calls
April 25, 2024 │ How to use modern machine learning tools to recognise animal calls
Porting and hacking Llama2.c for fun and profit
November 26, 2023 │ A devlog of my experience porting and hacking llama2.c
How LLMs work
November 26, 2023 │ A deep-dive into the specifics of LLMs, via llama2.c
Building LLM-driven agents
November 26, 2023 │ How ReAct-pattern agents work and how to build them yourself
Acing the coding challenge
December 25, 2020 │ Practical advice on writing a successful coding challenge
Acing the design interview
January 1, 2021 │ Practical advice on succeeding at a design interview
Blockchain explained simply
April 20, 2021 │ As simple an explanation of blockchain as possible
Building multiplayer games with socket.io and HTML5 Canvas
July 29, 2020 │ Building multiplayer games with socket.io and HTML5 Canvas
Blockchain for beginners
July 29, 2020 │ A brief explanation of how blockchains work for developers
What kind of work I want
September 6, 2021 │ A brief statement of values for recruiters or hiring managers
Keep incidents boring
April 20, 2021 │ The internet is full of war stories, but almost no incidents are like that
Learning incident response with problem sets
September 3, 2021 │ One idea for testing incident troubleshooting
Writing good technical explanations
December 25, 2020 │ How to explain complex technical concepts by starting with why
Avoiding worry driven development
September 11, 2020 │ Sofware dysfunction is more often motivated by worry than it is by a lack of technical skill