Mercor Machine Learning Engineer Interview Experience — Three Onsite Rounds, Kicked Out of the Meeting Room Mid-Interview

I finished interviewing at this company yesterday. My background is physics AI. I did competitive programming back in high school, and I figured I could cram a system design book for a few days and pass this onsite. Turns out I was still too green. The questions were pretty standard; it's on me for being too weak and wasting the opportunity. The interview was chaotic: in the middle of it, people kept knocking on the door saying they had booked the meeting room and kicking us out. The company looked exactly like what I imagined an SF startup would be, full of energy, with free drinks, fruit, and snacks.

1. Standard system design: building a workflow system.
2. Algorithm whiteboard discussion: finding a friend network, and how to call the API fewer times to get there. The API has some restrictions.
3. Hand-coding a LeetCode-style problem, screen-sharing to the meeting room display and writing it in my own IDE. Not an original LeetCode problem. The first part is super easy level: a line of cars parking, determine whether this line of cars can park, where each car has a preferred spot. The follow-up is basically standard DFS or backtracking. I was hacking at it purely on my high school knowledge, and I got this far before I ran out of time.

My guess is three no-hires. If I want to jump ship to MLE, I still need to grind some problems, even though this interview didn't feel like it had much to do with ML.