I was contacted by a recruiter. The interview process took about three months from initial contact to the final verdict.

The interview consisted of:

• An initial coding screen with two medium LeetCode questions.

The onsite consisted of four interviews:

• Two more interview rounds.
• A deep dive into deep learning.
• One ML systems design.

I had a couple of interviews, including three coding rounds, one deep dive into deep learning, and an ML design interview.

The coding round interview questions were LeetCode medium and hard. The interviewers were helpful and guided me, so I was able to solve them within a reasonable time. Be mindful of edge cases.

The ML design question is open-ended and hard to prepare for. You are asked to design a predictive system and told to focus on the machine learning side of things. Make sure to gather requirements (scale, scope, metrics to improve) before diving in. Talk about data collection. Consider what kind of data a company like Meta collects, how you could label it and use it, what other data sources you could use, how you would go about collecting them, and how you would label them. Be prepared to talk in detail about the ML model you think you should be using and defend your choice if needed.