April 5th, 2026

Articles of the week

This week I picked up these articles:

Learned

It's important for me to take my career as far as I can. I was happy to see that I have a lot of the habits in the first article, and I guess I'm also kind of good at using AI haha. Still worth sharing them since I enjoyed them.

7 simple habits of the best engineers I know

  • Be a great communicator. Communicate what you want, what you worked on.
  • Code fast, and slow. I love this, I think it's very important. I myself have a workflow/system that is related to this point that I developed a couple months ago and it's served me extremely well. I want to write about it soon.
  • Define/have coding standards, internalize the ones for your team.
  • Be ready to throw away the code, don't get attached to it.

The skill of using AI agents well

  • We (humans) are the bottleneck for AI, with our slow approvals when it needs approvals for permissions, our slow reviews. Identify how you are slowing down AI and try to improve it, but you won't always be able to.
  • I have yet to learn and use git trees, I always run Claude or Copilot in parallel but just by instructing it to do so, haven't literally run multiple paralle sessions, I shall give it a try soon.

Nobody is coming to save your career

  • Not everyone wants to grow their career, but if you are somebody who does, you need to be the one driving it.
  • Your manager most likely doesn't have "help x get to senior" in their list of top priorities, so you need to communicate it.
  • The moment you feel like you aren't learning, or you're getting comfortable, is when you need to take on a new project that scares you, ask for more work, etc.

Gergely Orosz on technical blogging

  • This man started writing because every developer he looked up to did it!

What is inference engineering

I really like the content Gergely puts out there, and I am subscribed to his Substack! If you want to stay up to date with the industry, if you want to learn all kinds of concepts, I recommend it tbh!

  • Inference is when a model takes an input and produces an output. Inference engineering is a way to optimize how models are used. This unlocks so many possibilities, you can grab a model and build an unique advantage by combining inference techniques and proprietary data and systems.
  • You can grab an open model and modify inference a ton of different ways, the one that caught my attention the most was quantization, which modifies the model's weights to improve latency.

I only read the first three parts and like a quarter of the quantization part, it's a technical heavy article and want to get through it during the week, but it's super interesting if anyone is curious about inference.