DEV Community: Sara H

I'm a Junior… but I'm a Team Lead!! 💫

Sara H — Mon, 08 Dec 2025 19:34:24 +0000

Yes, my official title is “Junior.”
But in today’s world — a world where LLMs are part of every step of the development workflow — every developer is actually a team lead.
Not of people… but of models.

The LLM is the “developer” working under me, and I’m the one managing it.
Exactly like a real team lead.

And that means:

1. Giving precise instructions 🎯

Just like a team lead guiding their developers, I must define clear,
detailed requirements with no room for ambiguity.
If I give a vague prompt — I’ll get a vague output.

2. Making sure it actually understood me 👌

Never assume.
I ask clarifying questions, refine the task, request examples, and confirm the model truly understands what I need before it starts generating
code.
(You are welcome to read my other post on this topic:
https://dev.to/sara_hajbi/how-i-work-with-llms-13ni)

3. Reviewing every line and every suggestion 🔍

An LLM is not an authority — it’s a tool.
I read the code, verify the logic, check the architecture, and look for bugs.
It’s my responsibility, just like a team lead reviewing a PR before merging.

4. Understanding 100% of what’s happening in the system 🤔

You can’t just throw a vague request at a model and wait for magic.
If you don’t understand what you’re running, you lose control quickly.

In today’s world, being “just a junior” isn’t enough.
You must be a developer who truly understands, not one who simply copies.

So what changed?

AI changed the rules.
A junior can’t remain “just a junior.”
You now have to act as:

a manager
a code reviewer
an architect
and the owner of the entire development process

…before writing a single line yourself.

Anyone who knows how to manage LLMs effectively is essentially a team lead of their models.
And those who don’t — fall behind.

What do you think?
Do you agree with me?

The One Question That Changed How I Work with LLMs!

Sara H — Mon, 08 Dec 2025 11:18:02 +0000

One of the most important things I’ve learned when working with large language models (like ChatGPT or Gemini) is not to assume they truly
understood me.

Yes, they will always say they understood.
But the truth? Not always. Sometimes they “invent” their own interpretation — and when you get to implementation, you realize the result is completely off track.

🎯 So here’s a small tip that completely transformed the quality of my results:
When I give a complex instruction, I don’t ask “Did you understand?” — instead I ask:

Do you have any more questions before we move on to implementation?

Was anything unclear to you? Don’t make assumptions — ask me.

And surprisingly — most of the time, the model actually asks really good questions.
Questions that force me to clarify myself, think through edge cases, and sometimes even realize what I haven’t fully decided yet.

📈 Since I started doing this, I’ve been getting outputs that are much more accurate, thoughtful, and clear.

🐳 Working with Docker? Don’t Let Your Disk Explode!

Sara H — Sun, 26 Oct 2025 20:45:15 +0000

If you work with Docker on a daily basis — you’ve probably seen it happen:
Everything runs smoothly, your builds succeed, containers are up…
And then suddenly — “No space left on device” or your computer slows down for no reason.

The truth? It’s not Docker’s fault — it’s yours 😉
Every time you run builds, create containers, or pull new images,
Docker keeps everything on your local disk:

Old images
Stopped containers
Unused volumes
Build cache layers
And it all piles up — fast.

How to Clean Up Docker?

Full Cleanup (Deletes Everything!)
docker system prune -a --volumes
What it does:

Removes all stopped containers
Removes all unused images
Removes all unused volumes
Clears build cache

💡 Tip: Add --dry-run first to preview what will be deleted.

Gentle Cleanup (Safe for Daily Use)
docker system prune
What it does:

Removes stopped containers, dangling images, and unused networks
Does not delete volumes
Safe to run regularly

Delete Images Only
docker image prune -a
Removes all images not currently used by any container.
Keeps your running containers intact.

Delete Volumes Only
docker volume prune
Cleans up volumes that aren’t connected to any running container.
Great for freeing up old persistent storage.

Delete Build Cache Only
docker builder prune
Removes build cache layers that accumulate over time.
Can easily free up several gigabytes of space.

If you work on multiple projects or rebuild images often,
make it a habit to run these cleanup commands once a week.
Good luck!💫