Project advice for newcomers

This is for juniors and software engineers interested in becoming AI Engineers. *Here, "AI Engineer" means a software engineer who builds LLM-based products.

You can showcase your skills with AI projects, but I don't care about your projects. Don't spam your résumé with multiple projects. In this vibecoding age, a project can be created with one prompt.

Why? You build the projects from scratch, implement everything yourself, and understand every single detail. There's no way to tell you vs the person who vibes it.

Just do one project. Build one product. Quality over quantity.

If it's an open-source contribution, even better. That's how you learn from legends. And I don’t mean you must have a pull request merged. You can contribute by helping people understand the project, and things like answering questions in community forums. The point is to understand a good portion of the project and show that you can work on a complex project.

Today, AI copilots can help you understand an entire project. Use them to teach yourself.

Find one framework that both the contributors and you are passionate about. e.g., llama.cpp, LLM frameworks. "Passion" is the keyword here. Passionate people will find ways to solve the hardest problems. Learn how people solve problems. Hang out in their community Discord. Ask questions, give answers. Be involved. Be willing to learn from everyone.

And, you can vibecode all you want. AI cannot write all the code. If you tell an experience where you have to take over AI and fix the code yourself, that's a strong signal. It's not just about AI helping you. It's about whether you have the skills to help AI. The core skill needed in engineering is the ability to solve problems, both simple and complex.

Disclaimer: This is not career advice. Past performance in engineering is not indicative of future performance. Skills may appreciate or depreciate over time. Always do your own due diligence before investing your nights and weekends.