TikTok Data Scientist Interview Questions
& Process
Real candidates share what happened, how many rounds they had,
and how the experience turned out.
Based on 189 interview experiences · FREE TO READ
Candidate interview experiences
First-hand accounts from people who interviewed at TikTok.
TikTok
Data Scientist Intern
Intern Average Neutral experience No offer 4 rounds
Interview process
Recruiter call Technical screen Onsite Onsite
Interview formats
Technical Coding Behavioral Case
I interviewed for a 2025 summer internship. The process involved 2 rounds plus an HR call and then a 4th round, but I ultimately didn't make it past the last one. The first round was about discussing a past data project, with follow-up questions on why I chose certain models. There was also one easy SQL question and one product-related question. Rounds two and four were entirely focused on product sense, specifically asking how to set up an A/B test and which metrics would be appropriate.
Confirmed questions6 questions
- Can you describe a data project you've worked on?
- Could you elaborate on the model choices made for that data project?
- Please provide a SQL query for a given scenario.
TikTok
Data Scientist
Mid Difficult Positive experience No offer 1 round
Interview process
Technical screen
Interview formats
Technical Coding
The interview process involved a technical screen where I was asked several questions. These included improving model performance on imbalanced data, designing A/B tests for ranking algorithms, handling sparse historical data for recommendations, detecting data leakage in time-series, and interpreting models with high accuracy but poor business KPIs.
Confirmed questions7 questions
- Can you share an instance where you enhanced model performance on imbalanced data, detailing your approach from data processing to evaluation and the rationale behind your choices?
- How would you set up an A/B test to evaluate two different ranking algorithms for a news feed?
- What metrics would you select and how would you manage bias in this A/B test?
TikTok
Data Scientist
Entry Average Negative experience No offer 2 rounds
Interview process
Recruiter call Technical screen Phone screen
Interview formats
Technical Behavioral
Just a heads up for anyone else applying, double check if you actually need to speak Chinese, even if the job description doesn't explicitly say so. The first red flag was no initial recruiter call. The recruiter couldn't speak English at all and emailed me instead. It was a bit weird not having a recruiter call to sort out logistics, but she said my resume looked good and set up the interview. The first interview was okay. The interviewer wasn't a native English speaker but could hold a conversation, and I think she'd be my manager. She was nice, and the questions were pretty standard for a data scientist. I passed this round. The second interview was bizarre - she spoke Chinese for 10 seconds, I told her I don't speak Chinese, then she spent 3 minutes trying to get the recruiter on the line. After that, she said she'd talk to the recruiter to figure things out and ended the interview in 5 minutes. Clearly, I didn't get the job because I couldn't speak Chinese.
Confirmed questions1 question
- What metrics would you propose for Tiktok Live?
TikTok Data Scientist Interview Questions
Quoted word for word from TikTok interview reports.
“Given a disease prevalence rate of 0.1%, a True Positive rate of 100%, and a False Positive rate of 5%, what is the probability that a person who tested positive is actually carrying the disease?”
Read reports →“Given tables detailing TikTok user sign-ups and confirmations via email and text, where users sign up with email and get a text to confirm, write a query to find user IDs of those who signed up but didn't confirm on day one, yet confirmed on day two.”
Read reports →“Could you write a SQL query to determine the average visiting times over the last 30 days?”
Read reports →“For SQL1, how would you examine video upload event statistics from non-US countries in a given time frame, considering SELECT, WHERE, GROUP BY, ORDER BY, and case-insensitivity for the country field using UPPER?”
Read reports →“For SQL3, how would you count today's viewing events from non-US countries and rank them by the average fraudulent sessions in the last 30 days, using CTEs, RANK() and OVER() window functions, and date offsets?”
Read report →“For SQL2, how do you count daily active users in the US on a specific date, filtering out high-risk users, and ensuring accurate counts with JOIN and COUNT(DISTINCT user_id)?”
Read report →“How do you address issues caused by imbalanced data, for example, using SMOTE, and why is this technique significant?”
Read report →“What kind of metric would you choose to measure whether the answer provided by AI in the Tiktok app is useful to the users?”
Read report →“If multiple experiments are running concurrently, how can you determine if the observed results are attributable to a specific experiment?”
Read report →Formats, difficulty and experience
Across all 189 TikTok interview reports.