Why you should work for a top tier tech company
I joined a FAANG company 1.5 years ago after working mostly in startups and one medium sized company. I wish I’d joined way sooner because I’d have millions more in my bank account. But besides the…
RAW.works - RLMs are the new reasoning models
Codex recently coded a browser from scratch. It took a week, but the AI worked solo, with no human in the loop. Could a human parse that code and truly understand it? The bigger question is: does it matter? Have you ever audited your binary’s assembly code? Yo…
Three Time-to-Power Strategies that Failed in 2025
We’ve learned a lot in the last year about what won’t work for accelerating time-to-power for large data center projects.
The Missing Bundler Features
Over the last few months, there has been a lot of talk about making Bundler faster, both by improving it directly, or by reimplementing it in another language, and while it may surprise some, that didn’t excite me much.
The sycophancy problem, the compute crunch, and why coding agents will not stay cheap.
(Part 2) The inference side of transformers, along with systems tricks that make production LLMs fast
Strategies that worked for me when I left a salaried job to consult on my own.
Apple announced Tim Cook will become executive chairman of Apple’s board of directors and John Ternus will become Apple’s next CEO.
What is Canton Network (and why should you care)?
Canton Network is a layer-1 blockchain designed for regulated financial markets, addressing privacy, regulatory compliance, and interoperability without centralized control. Unlike typical public b…
When “Right vs Wrong” Isn’t Really About Truth — A Thought on Bias, Power, and Decision Making
By: Neeraj R One day, while attending a class on human bias, a simple thought experiment turned into something much deeper. It started with a very ordinary scenario. Imagine a group of 10 friends p…
Decision Graphs
About eight years ago, I wrote a little blog post on dynamic decision trees. If you have the time, go back and read that and the follow up posts on the subject. If not, I’ll summarize. Instea…
Connecting in Narratives
Missing a flight to my cruise can teach us how the right narratives invite others on the journey with us.
When you take a close-up photo of a small object, there tends to be a small portion of the subject in focus, with the rest blurred out. The range of distance that's acceptably in focus is the depth of field; with macro photography it tends to be quite narrow.
Bricked hardware, ads, and control. The Amazon Kindle is no longer built for readers. Here’s why I’m done with it.
A UX question from a restauranteur
I’m Back. Here’s my life story in a nutshell
I haven’t posted anything here in almost 3 years. The main reason is that because after a grueling job search in 2024, I lucked into joining a Big Tech / FAANG company as a senior software engineer…
Looking back on Stripe’s payment API migration
This 2020 post walks through the first and second iteration of Stripe’s payment API. The original API was designed for credit card charges and was delightfully simple. You could move money with a s…
The Ultimate Question: What Does the Endgame Look Like? - Pawel Brodzinski on Leadership in Technology
Asking "what does the endgame look like?" will show how many predictions can't hold or what other second-order consequences we will face.
Testing open-weight models against unpublished vulnerabilities in real products.
Language Modeling Without Neural Networks
Generating Shakespeare has become the “Hello World” of language models.1 Recently, I’ve been messing with alternative language models and came across unbounded …
A practical look at deploying Claude across an organization: skills, plugins, governance, and real-world challenges from a 23-person team.
Hauleth's blog - Scotty, I need warp speed in three minutes
My journey into optimising Elixir codebase of Ultravisor (my fork of Supabase's Supavisor). This story is not about a goal, but ~~friends~~ optimisations we met along the way.
Stop shipping tables. Build data products with clear contracts, real owners, lifecycle discipline, and observability so teams can trust data.
Introducing ggsql, a grammar of graphics for SQL that lets you describe visualizations directly inside SQL queries.
We analyzed 59,000+ daily records from 256 sauna users to quantify what saunas actually do to your body. Using paired t-tests on same-day wearable data, we foun
ME2-Writeup/README.md at master · coremaze/ME2-Writeup
Reconstructing a Dead USB Protocol: A Handheld's Secrets Unlocked by a Hot Knife -- A multi-disciplinary journey to reviving a forgotten USB interface - coremaze/ME2-Writeup
Kimi K2.6 advances open-source coding, featuring long-horizon coding, coding-driven design, agent swarms, proactive agents, and the Claw Groups research preview.
Qwen Studio
QWEN STUDIO DISCORD Following the release of Qwen3.6-Plus, we are sharing an early preview of our next proprietary model: Qwen3.6-Max-Preview. Compared to Qwen3.6-Plus, this preview release brings stronger world knowledge and instruction following, along with …
A couple of years ago, I wrote that The Builder pattern is a finite state machine!. A state machine consists of states and transitions between them. As a developer, I want to make illegal states unrepresentable, i.e., users of my API can’t create non-existent …
Atlassian Enables Default Data Collection to Train AI
Atlassian will begin collecting customer metadata and in-app content from **Jira**, **Confluence**, and other cloud products by default on **August 17, 2026**, to train its AI offerings including `Rovo` and `Rovo Dev`. The change affects roughly **300,000** cu…
EU to force replaceable batteries in phones and tablets from 2027
SMARTPHONES and tablets sold in Europe will have to feature replaceable batteries starting next year, according to EU rules, amid efforts to slash
Comment Tesla a caché des accidents fatals pour continuer à tester la conduite autonome sur les routes | RTS
Une fuite de données révèle que Tesla a dissimulé des milliers d'incidents liés à sa conduite autonome. Certains accidents ont été fatals. Un premier verdict condamne le constructeur à verser 243 millions de dollars aux victimes. Une enquête diffusée dans Temp…
Inside GitHub's Fake Star Economy
Six million fake stars, $0.06 per click, and a VC funding pipeline that treats GitHub popularity as proof of traction. We ran our own analysis on 20 repos and found the fingerprints.
I spend a lot of time negotiating this in the software world: And if you're wondering why this happens, it's normally because: 1. people aren't talking to people 2. people aren't listening So lots of designers and product people have leapt onto 1, basical…