This as-told-to essay is based on a conversation with Prakhar Agarwal, an applied researcher at Meta Superintelligence Labs. The following has been edited for length and clarity. Business Insider has verified his employment and academic history.
—
**My Journey into AI: From Apple to Meta Superintelligence Labs**
I started my career at Apple in 2020, spending five years there before moving to OpenAI as part of the OpenAI API team. This summer, I joined Meta Superintelligence Labs, following many others who made the same shift.
At the time I applied to Apple, I was in graduate school at the University of Washington, specializing in machine learning. Later on, I didn’t have to apply explicitly to OpenAI, Meta, or several other companies—they began reaching out to me directly.
Experience plays a huge role in landing these positions. The number of openings at top AI labs is quite small, so naturally, they tend to hire experienced candidates.
—
**High Autonomy Roles in Top AI Labs**
The roles I’ve held involve a high degree of autonomy. Unlike traditional corporate setups with clear hierarchies, these positions require you to identify gaps in current technology and independently work on solutions.
You’re responsible for prioritizing which problems to address, given your limited time and resources. Once you’re in, you’re essentially thrown in the deep end—defining your own problems and devising your own solutions.
At OpenAI and Meta, the focus is heavily on hiring smart people. It’s expected that you come in knowing what needs to be done, rather than being told.
—
**Interviewing at a Top AI Lab**
The interviews at these labs test a couple of key skills. First, you need to understand the required nomenclature and foundational concepts around Large Language Models (LLMs). Writing code is still part of the process, but it is much more involved and directly connected to the actual work.
Secondly, interviewers want to see if you can operate effectively in ambiguous domains. Given an abstract problem, can you concretize it into a workable, metric-driven solution?
Having a Ph.D. certainly helps, as it demonstrates your ability to handle complex, abstract problems. However, practical experience—whether at a startup or building integral software components—can be equally valuable in getting your résumé noticed.
I recommend getting hands-on experience—actually working on problems and solutions. This practical work builds your skillset and intuition, teaching you not only what to do but what won’t work. That intuition will help you stand out during interviews.
—
**Top Tips for Getting Hired**
At a minimum, ensure your theoretical understanding of AI is solid. Learn the necessary nomenclature specific to your job.
Use AI models extensively. By working with them, you’ll better understand their strengths and weaknesses—something many candidates overlook.
One of the most sought-after abilities is to identify gaps in AI models. Ask yourself, “What is a gap that needs addressing in the next version of LLaMA?” Once identified, can you quantify this gap with a metric?
It’s also important to demonstrate awareness of trends—predicting the capabilities that models might have three to six months down the line.
—
**The Value of High-Bandwidth Communication**
Top AI companies prioritize high-bandwidth communication. Problem-solving happens at a much faster pace compared to traditional Big Tech firms, where you might spend a week preparing presentations.
Here, you often jump straight into a whiteboard session or a small meeting before going off to work independently.
Work conversations are usually one-on-ones or small groups, so being able to clearly articulate problems and gaps to peers and supervisors is essential.
—
**How to Actually Learn AI**
One thing I’ve noticed about AI communities is their openness to sharing ideas and feedback. If you get stuck, don’t hesitate to reach out to people on Twitter or LinkedIn—most are willing to help.
Classroom coursework can feel outdated in this rapidly evolving field. To keep pace, don’t rely solely on traditional courses or textbooks written five or ten years ago.
Consume knowledge from wherever it’s available: blog posts, YouTube videos, Twitter threads, or direct conversations with professionals.
Start following thought leaders who actively share insights in AI. You might not grasp everything immediately, but over time, you’ll begin to pick it up naturally.
—
Prakhar Agarwal’s insights provide a valuable window into the realities of working at top AI research labs and offer practical advice for those aiming to break into this exciting field.
https://www.businessinsider.com/openai-meta-superintelligence-labs-tips-getting-hired-phd-llm-interview-2025-10