TikTok Data Scientist interview questions

111 Data Scientists have shared notes from their TikTok interviews, with the questions they were asked in each round. That is 40% of the 279 TikTok interview notes on Screna AI.

111 Data Scientist interview notes

TikTok·Data Scientist·Technical Phone Screen

Jul 2026
TikTok data scientist interview that went deep into experimentation statistics. One big multi-part question covering basically every corner of A/B testing you can imagine. Left feeling like I'd either nailed it or completely embarrassed myself, still not sure which.
  • You ran an A/B test on a video length change and got a non-significant result on your primary metric. Baseline completion rate is 22% and the product team wants at least a 5% relative lift to matter. Walk through how you'd calculate the required per-arm sample size for 80% power at a two-sided alpha of 0.05, and explain your assumptions.
  • How would you apply variance reduction techniques like CUPED using pre-period completion data, and if your covariate explains 25% of the outcome variance, what's the practical effect on your test's power?
  • Given the non-significant result, should you keep collecting data, stop the experiment, or switch to a Bayesian approach with a utility-based stopping threshold? Defend your choice.
  • If you're tracking multiple related outcomes alongside your primary metric, how do you handle the multiplicity problem and what corrections would you apply?
  • How would you design a sequential testing approach, like an O'Brien-Fleming spending function, that allows you to peek at results early without inflating your Type I error rate?

“This part I actually felt okay about.”

View Post

TikTok·Data Scientist·Recruiter / HR Screen

Jul 2026
Screener for a Data Scientist role at TikTok that opened with a surprisingly basic finance question, felt more like a vibe check than a technical round.
  • What does 'credit' mean in a financial context, and why does it matter to both consumers and banks?

“Wasn't expecting this at all for a data science screen.”

View Post

TikTok·Data Scientist·Hiring Manager Screen

Jul 2026
TikTok data scientist interview focused entirely on e-commerce risk control, which makes sense given the role but still felt pretty niche if you haven't lived in that space. Four questions, all with some teeth to them, ranging from the soft intro pitch to a genuinely tricky prioritization scenario.
  • Give a 30 to 60 second overview of your background in e-commerce risk control, then explain why you want to join this specific risk team and why now.
  • What are your top three goals for your first 30, 90, and 180 days in this role, and which specific metrics would you use to measure whether you're succeeding?
  • Tell me about a time you pushed back on a risk policy or threshold. What data did you bring to the conversation, what trade-offs did you put numbers on, and what happened?
  • If you were handed competing projects simultaneously, how would you prioritize them given a small analyst team and only about 10% of engineering bandwidth available?

“The intro part is fine, everyone has a spiel.”

View Post

TikTok·Data Scientist·Technical Phone Screen

Jul 2026
TikTok data scientist interview focused heavily on e-commerce advertising analytics, covering everything from metric design to A/B testing to diagnosing revenue drops. Pretty case-heavy, less coding than I expected.
  • Walk me through one of your resume projects: what success metrics did you define, how did you measure them, and what did your A/B experiment design look like?
  • The business wants to grow merchant advertising revenue or merchant GMV. How would you approach diagnosing the current state and recommending actions?
  • Ad ROI has been declining recently. How would you investigate why, and what fixes would you propose?

“I picked a project I thought was solid but fumbled the experiment design part a bit.”

View Post

TikTok·Data Scientist·Technical Phone Screen

Jul 2026
TikTok data scientist interview, technical phone screen or onsite SQL round. One meaty query question that looked straightforward until I started counting all the edge cases they wanted handled.
  • Using ad delivery logs, write a SQL query to find advertisers whose CTR dropped by at least 20% in the most recent 7-day window compared to the prior 7-day window. Exclude flagged suspicious ads, enforce a minimum impression threshold of 10,000 per window per advertiser, handle missing days as zero, guard against division-by-zero and integer division, and return advertiser_id, both CTRs, and the relative drop ordered by drop descending.

“This one had way more moving parts than it looked.”

View Post

TikTok·Data Scientist·Technical Phone Screen

Jul 2026
TikTok data scientist interview with a stats-heavy question about comparing creator behavior across cohorts. Pretty technical, felt like they wanted to see if you actually know when to use which test rather than just naming tools.
  • After computing cohort-level posting metrics, how would you determine whether posting patterns differ across cohorts? Walk through the metrics, visualizations, and statistical tests you'd use, and explain what would drive your final conclusion.

“I started with the visual layer, retention curves overlaid per cohort, average posts per user over time, that kind of thing.”

View Post

TikTok·Data Scientist·Onsite - Behavioral / Leadership

Jul 2026
TikTok data scientist interview with a meaty behavioral question that was way more structured than I expected. They really wanted you to have lived this, not just theorized it.
  • Tell me about a time you disagreed with a partner team over a product direction and still drove the work end-to-end. Walk through the decision you proposed, the metrics you used to define success (baseline, expected lift, acceptable regression), how you structured the rollout or experiment, how you handled cross-timezone coordination, and one trade-off you consciously accepted.

“This question is doing a lot of heavy lifting at once and I underestimated it.”

View Post

TikTok·Data Scientist·Technical Phone Screen

Jul 2026
TikTok Data Scientist interview that went deep into classical AI search algorithms. More graph theory than I expected for a DS role, but looking back it makes sense given how much recommendation and ranking work they do.
  • Given a weighted graph with nodes S, A, B, C, D, G and specific edge weights, trace the node expansion order and final path found by BFS (treating edges as unweighted), uniform-cost search, and A* with the provided heuristic values. Break ties alphabetically.
  • Prove whether the given heuristic h is admissible and consistent for the provided graph.
  • Compare the time and space complexity of DFS, BFS, UCS, and A* in terms of branching factor b and depth d. Describe a scenario where DFS wins on memory but fails to give an optimal solution.
  • Under what condition does A* reduce to Dijkstra's algorithm? Illustrate using the heuristic values provided.

“This is where I spent most of my mental energy.”

View Post

TikTok·Data Scientist·Technical Phone Screen

Jul 2026
TikTok Data Scientist SQL round, one question but it was a beast. The kind of problem where you think you understand it after the first read and then realize you missed three edge cases by the second.
  • Write standard SQL (no procedural code) to compute each user's total valid usage minutes for a specific date. You need to clip activity intervals to that day's boundaries, merge overlapping or adjacent intervals per user, then check whether any gap between consecutive merged intervals exceeds 60 minutes. If any such gap exists, return NULL for that user's usage. Otherwise return the total merged duration in whole minutes (floor).

“The clipping part I got pretty quickly, GREATEST and LEAST to bound start and end to the day window.”

View Post

TikTok·Data Scientist·Technical Phone Screen

Jul 2026
TikTok data scientist interview that went deep into causal inference territory. The main question was a sprawling DiD design problem covering staggered rollouts, estimator choice, inference, and reconciliation with PSM. Felt more like a methods exam than a conversation.
  • You're running a staggered rollout across 50 regions over roughly 10 weeks with weekly revenue per user as your outcome. Design a difference-in-differences analysis that holds up under heterogeneous treatment effects. Walk through when two-way fixed effects is biased, what it actually identifies, and how you'd implement a Callaway-Sant'Anna or Sun-Abraham estimator including group-time ATTs and aggregation weights. Also cover the event-study setup, pre-trend testing with a joint F-test, inference under few clusters, and how you'd reconcile your DiD results with a parallel PSM analysis.

“This was basically six questions stapled together and I definitely felt the seams.”

View Post