Datadog Data Scientist interviews mix a recruiter screen with technical rounds that vary by interviewer—expect SQL gap-and-island problems, A/B testing case studies, ML fundamentals, or LeetCode-style coding, followed by a project-focused hiring manager round.
$149K
Avg. Base Comp
$183K
Avg. Total Comp
2-5 rounds
Typical Rounds
Not reported
Process Length
Datadog's Data Scientist interviews typically open with a recruiter screen before branching into a handful of different technical formats, and which one you get seems to depend heavily on the interviewer or team. One candidate described the toughest SQL question of their entire nine-month job search here: grouping logins within a 30-minute window into session IDs, a classic gap-and-island SQL problem that goes well beyond standard joins and aggregations. Another candidate's process centered almost entirely on experimentation: designing an A/B test end-to-end, choosing primary, secondary, and guardrail metrics, and connecting technical choices to business outcomes rather than reciting formulas. Other reports describe a more traditional fundamentals round covering anomaly detection on time series, OLS regression, and how to model a dataset too large to fit in memory, or a probability-and-statistics-heavy discussion with comparatively little coding.
Coding shows up too, usually as LeetCode easy-to-medium problems in Python or on Coderpad, sometimes as a quick warmup before a harder round with a senior engineer. Later stages can combine live data analysis with ML system design, and the final hiring manager conversation can focus heavily on defending a past project's technical trade-offs. One candidate found this round unexpectedly adversarial and declined the resulting offer over it, while an intern candidate's final round with the data science team felt more like a collaborative discussion of past projects and technical decisions. Because the technical emphasis varies so much by interviewer, it's worth brushing up broadly across SQL, statistics, A/B testing, and ML fundamentals rather than betting on one format.
Synthesized from 7 candidate reports by our editorial team.
Had an interview recently?
Share your experience. Unlock the full guide.
Real interview reports from people who went through the Datadog process.
Share your own interview experience to unlock all reports, or subscribe for full access.
Sourced from candidate reports and verified by our team.
Topics based on recent interview experiences.
Featured question at Datadog
Minimum Change
| Question | |
|---|---|
| Network Experiment Design | |
| The Brackets Problem | |
| Basic Regex | |
| Hurdles In Data Projects | |
| Trial User Segmentation | |
| Production Rollout Challenges | |
| Client Solution Pushback | |
| Newsfeed Model | |
| Data Cleaning Experiences | |
| Docs Metrics | |
| LRU Cache 1 | |
| Empty Neighborhoods | |
| 2nd Highest Salary | |
| Top Three Salaries | |
| Rolling Bank Transactions | |
| Merge Sorted Lists | |
| Customer Orders | |
| Comments Histogram | |
| Closest SAT Scores | |
| Subscription Overlap | |
| First to Six | |
| Monthly Customer Report | |
| Experiment Validity | |
| Download Facts | |
| Upsell Transactions | |
| First Touch Attribution | |
| Employee Salaries (ETL Error) | |
| Random SQL Sample | |
| Compute Deviation |
Synthesized from candidate reports. Individual experiences may vary.
Candidates consistently report starting with a recruiter or HR phone call covering background, motivation for joining Datadog, and what they are looking for in the role. This stage is conversational rather than technical.
This round varies noticeably by candidate. One report centered on a gap-and-island SQL problem requiring session/streak grouping of timestamped events. Another was a probability-and-statistics-heavy discussion with Python coding. A third tested anomaly detection, OLS regression, and memory-efficient modeling for datasets too large to fit in memory. Because the focus differs this much, preparing broadly across these areas is the safer bet.
Several candidates describe a dedicated coding stage with LeetCode easy-to-medium problems, sometimes as a quick warmup before a harder medium-difficulty problem with a senior engineer, and sometimes as a full hour on Coderpad. Interviewers are typically described as collaborative, focused on problem-solving approach rather than pure speed.
One candidate's technical case study focused entirely on experiment design: defining the problem, choosing primary, secondary, and guardrail metrics, setting control/treatment groups, and discussing sample size and power. Other candidates instead describe a live data analysis session, in one case combined with ML system design and in another using observability metrics and latency problems.
The closing stage often asks candidates to present and defend a past project or discuss experience and values. One candidate presented a complex project blending ML and software engineering, found the hiring manager unusually adversarial, and declined the eventual offer afterward. An intern candidate's final round with the data science team was a collaborative walkthrough of past projects and technical decisions. Experiences here appear to depend heavily on the specific interviewer.
