that's quite some swap usage

Every new unit of business logic, whether it’s an AWS Lambda function, a scheduled job, or a background worker behaves like a microservice in practice, requiring its own monitoring, scaling, debugging, and updates. The cost of context switch adds up quickly.

I got to experience this once, and that was the best DX I've ever seen. Not relying on available context and memory and still being able to refactor changes you.

If the number of microservices outpaces the team’s ability to maintain and evolve them effectively, complexity becomes a bottleneck, not a benefit.

In software engineering, there’s no LLM that can replace true ownership and accountability.

The best way to write an AWS Lambda is to not write one. It's a half baked microservice with added bonus of split observability.

When you look beyond the C.R.E.A.M should you find the moment of truth.

Perplexity's crawler out there being sneaky like a ninja. IETF RFC who?

"Type discipline isn't a limitation, it's a long term investment."

Python backend systems without type hints and absolute imports are a breeding ground for bugs. The further you move away from them, the less maintainable your code becomes and eventually, it collapses under its own weight.

AWS CodeCommit is performance wise the worst version control service ever. It's a mess to have repositories in it. Even if its closed for new customers, old ones are still using it.

one recent one is from @kepano stephango.com/ramblings

Love reading short blog posts that dive into a concept or share fresh ideas through personal/team stories.

some people just have a knack for distilling technical information to the highest level of signal possible wow

As the author of Windows Task Manager, I'm a huge fan of htop on Linux and NetBSD. Here it is running on my VAX 4000! The original author of htop is Hisham H. Muhammad —a developer who created the tool in 2004, hence the name “htop”.

World religions in shambles as Anthropic researchers reveal that Good and Evil are nothing more than vectors in latent space

😶

some of anthropic rugpulls so far: > windsurf no access to claude 4 > plus not getting opus 4 in claude code > 1.58-bit quantized models during daytime > max plans limits cut in half 2 weeks ago, no comms > weekly limits without concrete numbers > 5x/20x plans being actually 3x/8x of plus > cutting off openai api access what an absolutely horrible company

The title of the post could be better though.