I’ve built event-driven systems processing 1M+ events per month at startups. One thing that always caused headaches: Validation was left to downstream consumers. In Kafka, any data can enter a topic. Every consumer has to do its own validation. Miss one rule, and you end up

Good thing we sunsetted ServerlessQ - Your Serverless Message Queue some time back (for those not remembering: it was an abstraction over SQS/Lambda/DDB for Vercel users). I think Vercel Queues can have insane potential. The biggest bottleneck for many fullstack devs who are coming from a frontend side is,

I am trying to migrate from AI SDK v4 to v5. I thought I would be done and tested my changes... All of a sudden the responses do not match my defined zod schema anymore. Why?! 🤔

I like the concept of a „luck surface area“. It’s in your hands to increase that area even though it stays a game of probabilities…

- I only used Curser tab completion - I wanted to learn something new - I discovered lazy.vim - I have a split keyboard 👉🏼 Switching to lazy.vim 😎

My two favorite git aliases: wip = !git add --all && git commit -m "WIP" --no-verify uncommit = reset --soft HEAD^ I use 'wip' after each Claude Code completion. In the end, I use 'uncommit' to have all files staged and create a commit.

Is Omarchy Linux to Arch what LazyVim is to nvim? 🤔

This is pretty cool 🙌🏼

I prefer to reduce transparency in iOS 26. The glass effect sucks and makes things hard to read.

Is there a Neovim, world leader in plugin deliveries plugin that adds AI autocomplete (tab complete) that can use my existing Claude Max plan? 🤔

What a genius move from Anthropic to register CLAUDE.md. Every time someone refers to the file, it actually links to their docs 👏🏼

This is NOT a good way of doing it. Everything that doesn’t have to be managed by React shouldn’t be managed by React. Just create a function outside of React for formatting. No need to useMemo (which falls apart here anyways since it is an object and you can’t guarantee a

Love this! This is exactly how I think. I couldn't have said it better, though.

I am using React for quite some years (since hooks came out). I feel that it got more complicated over time and I got worse using it. It seems way to complicated. I feel the same about Next 😕

This is already the case. Skilled developers that also know how to leverage AI become 10x. Those who suck at programming OR don’t know how to use AI will become 0.1x (compared to the new baseline).

I have heard good things about Fraud Falcon 🤔 😜

Most people I know worry about their AWS accounts' security. With no spending cap, this is a valid concern. Today, with LLMs & MCP servers, security becomes even harder. I love using MCP servers for AWS tasks, especially with the large set of official ones provided by AWS.