I applied through a recruiter. I interviewed at TikTok
Interview
1) Coding: One medium-hard problem 2) System-Design 2 round: ML fundamental questions(what is transformer, what is attention model). recommend system / their own team related system design questions.
Stop at system design. background is not fit.
I applied through a recruiter. The process took 2 months. I interviewed at TikTok in Jul 2025
Interview
Deeply scrutinize resumes + ask for extensive LLM knowledge + moderate Leetcode skills
Resume questions are very detailed, requiring clear explanations and asking about many related topics. LLM knowledge is about Transformers, such as explaining QKV calculations and calculating flops.
Leetcode is generally not difficult, focusing on binary search, trees, and graphs.
Interview questions [1]
Question 1
What is the formula to compute the attention score given by Q, K, V
I applied online. The process took 2 months. I interviewed at TikTok in Sep 2025
Interview
I applied online through a job board site for the San Jose office. The process took a little over two months. I had 6 rounds of interviews.
First round: First 45min ask about PhD research, last 15 min medium leetcode question (longest increasing subsequence problem).
Second round: First 30min ask about PhD research. Last 30min ML design question, "How to filter out gun violence from TikTok videos?"
Third round: First 30min ask about PhD research. Last 30min ML design question, "How to determine whether a hashtag is relevant to a video or not?"
Fourth Round: HR interview lasting 30sih minutes, basic behavioral questions like "how do you handle harsh criticism?" The interviewer mentioned that they had a tight headcount and that they would get back to me within a week or two.
I got contacted by two other HR members who said my profile got noticed by two other leaders and I would have two additional interview rounds. My understanding is that they filled the role for the original team they were thinking of matching me with (safety and trust team) and the search team was interested in now interviewing me.
5th round interview: TLM of the search team was interviewer. First 25 minutes describe PhD research, last 45 minutes ML design question, "How would you determine from a large set of sparse features which ones are the most important to use for predicting CTR on a TikTok video?"
6th Round Interview: Head of Search, USDS was interviewer. This was a "behavioral round" however, it was more about trying to sell myself why I should work on the search team. One question that was asked: "How would you handle a deadline that you couldn't make?"
Overall the interview experience took an incredibly long time and it was hard to understand why I didn't match to either team at the end. That being said each interviewer was polite, clear and helpful during the interviews.