I'm a Stanford professor and AI startup cofounder. Here's how to get a job at an AI company.

• AI job seekers should build real projects using public data sets, said Stanford professor Jure Leskovec.
• Also a startup cofounder, he said adaptability and curiosity are crucial as AI evolves rapidly.
• Communication and empathy matter as much as knowing how to code, Leskovec added.

This as-told-to essay is based on a conversation and written commentary from Jure Leskovec, a computer-science professor at Stanford University and cofounder of Kumo, a maker of AI tools for predicting business outcomes from company data. This story has been edited for length and clarity.

If you want to work in AI, you need to show that you can actually do the work. Launch real projects using public datasets, deploy a demo, post your work on GitHub, or write about it on a blog.

Participate in hackathons — they're a fantastic way to demonstrate initiative and teamwork in a short time. We organize hackathons ourselves and are often impressed by what participants produce. It's concrete proof of what you can do.

Even if you fail, you're showing that you're curious and proactive. By your second or third project or hackathon, you'll have gained valuable experience.

We recently hired someone who stood out because he built a generative AI tool for analyzing customer purchase data. It showed ambition, curiosity, and problem-solving, which are qualities we really value.

Curiosity and flexibility matter

My second recommendation is to show adaptability — that you're the kind of person who is always experimenting with new tools, and that you can learn quickly. This is essential because AI is evolving at a pace that surprises even those of us who work in the field every day.

The best job candidates have taught themselves frameworks like PyTorch, JAX, or LLM tooling, and they stay current on areas like GenAI, multimodal models, diffusion, and reinforcement learning. Curiosity and flexibility matter more than having a fixed set of skills, because the skills in demand today may look very different tomorrow.

A top school or credential might get your application looked at, but it won't get you hired. We look for people who build things, who are adaptable and curious. In job interviews, we can tell if someone is just trying to map new ideas to what they learned in school versus truly engaging with what's new.

There's no playbook for AI. We're writing it right now. I always value it when my students bring me solutions that haven't been tried before, even if they're wrong. We're still at the experimental stage of AI in many ways, and there isn't always a clear textbook answer.

Sharpen your thinking

At Kumo, we conduct several interviews to see an applicant's full thought process. We pay close attention to how they approach problems and often value their reasoning as much as their final answer — if not more.

It may sound simple to say, "think outside the box," but it is more critical now than ever. Who knows? Your idea today could become the standard tomorrow.

I encourage people to sharpen their thinking by questioning assumptions, trying to solve problems without relying on familiar tools, and deliberately exposing themselves to new domains. Practice brainstorming multiple answers to the same problem, even the ones that seem impractical at first. Over time, these habits train you to see possibilities others overlook.

One last piece of advice: Don't forget to be human. Technical skills aren't everything. I look for people who can communicate clearly, work well in teams, and think carefully about the ethical and social implications of what they're building. Collaboration, empathy, and awareness of bias matter just as much as knowing how to code.