Snowflake data science candidates report a mix of experimentation, ML theory, SQL and Python, and case-study or project discussions. Here is what four reports describe and how to prepare for it.
$159K
Avg. Base Comp
$241K
Avg. Total Comp
Not reported
Typical Rounds
Not reported
Process Length
Candidates describe Snowflake data science interviews as broader than a coding screen. The Data Scientist report opened with an A/B test design for query latency, where follow-ups went deep on unit of analysis and variance reduction. The same candidate then met a supervised-learning theory interviewer and a research-background discussion on stochastic processes and time series. There was also a system design-style conversation about matching data across systems. The Python and SQL questions were described as easy LeetCode-level.
Two internship candidates describe a different front door. One report is a HackerRank assessment with multiple-choice, SQL, Python and a housing-price prediction task. The other is a longer assessment where a regression model had to hit an accuracy threshold on a messy dataset. One of them then got an ML-theory interview covering activation functions, transformers and transfer learning, with no coding despite what the recruiter had said. A Senior Data Scientist reported seven rounds. These covered a project walkthrough, model selection (F1 versus ROC-AUC, correlated features, overfitting, maximum likelihood in logistic regression) and a final case study. That candidate also found the scientist-versus-engineer scope unclear.
All four reports ended without an offer, so treat them as preparation hints rather than a fixed loop. Practice experiment design and SQL, and be ready to explain your modeling choices out loud. Ask your recruiter early what each round will cover and what the role's scope is.
Synthesized from 4 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 Snowflake process.
The first round caught me off guard because it was much more about experimentation than I expected for a data science role. I was asked to design an A/B test for query latency, and the discussion went pretty deep into unit analysis and variance reduction. I had actually just studied a very similar problem, so the content itself was manageable, but the interviewer was rushed and kept interrupting, which made it harder to lay out my reasoning cleanly. The follow-ups felt a little chaotic, and I left that round feeling like I never really got to show the full picture of how I think through an experiment.
The rest of the process was a mix of technical and more conversational rounds. There was a theoretical machine learning interview centered on supervised learning, and another round that dug into my research background, especially stochastic processes and time-series modeling. I also had a system design-style discussion focused on data matching across different systems, which felt tied to the interviewer’s own work. On the coding side, the Python and SQL questions were straightforward, more like easy LeetCode-level problems than anything deeply algorithmic. The final conversations with the hiring manager and other leadership/XFN folks were fairly informal, but the overall process still felt a bit disorganized. In the end I didn’t get an offer, and the biggest takeaway for me was to be ready for a broad mix of experimentation, ML theory, research depth, and practical data design questions rather than expecting a purely coding-heavy loop.
Prep tip from this candidate
Be ready to walk through an A/B test for query latency in detail, especially unit of analysis and variance reduction. I’d also review supervised learning fundamentals, time-series/stochastic-process concepts, and practice explaining a data-matching system design clearly and concisely.
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 Snowflake
Select the 2nd highest salary in the engineering department
| Question | |
|---|---|
| Comments Histogram | |
| Random SQL Sample | |
| The Brackets Problem | |
| Level Of Rain Water In 2D Terrain | |
| Basic Regex | |
| Marketing Channel Metrics | |
| Cumulative Sales By Product | |
| Sample Time Series | |
| Feature Store Integration | |
| Average Unique Counts | |
| Merge N Sorted Lists | |
| Log Anomaly Detection Model | |
| D2C Socks e-Commerce | |
| Minimum Days for Scheduling All Meetings | |
| Shortest Path Algorithms | |
| Client Solution Pushback | |
| Empty Neighborhoods | |
| Merge Sorted Lists | |
| Closest SAT Scores | |
| Subscription Overlap | |
| Rolling Bank Transactions | |
| Customer Orders | |
| Top Three Salaries | |
| Upsell Transactions | |
| First to Six | |
| Monthly Customer Report | |
| Experiment Validity | |
| Download Facts | |
| First Touch Attribution |
Synthesized from candidate reports. Individual experiences may vary.
Both internship candidates report starting with a HackerRank assessment. One describes five questions in 90 minutes: two multiple-choice, one SQL, one Python list problem and a house-price prediction task. The other describes a longer assessment with a regression model that had to meet an accuracy threshold. Full-time loops may differ, so confirm the format with your recruiter.
The senior candidate reports the process starting with a hiring manager call, with a recruiter joining to coordinate timing and location. That candidate also found the data scientist versus data engineer scope unclear, so it may help to ask early what the role emphasizes.
Candidates report technical rounds covering A/B test design for query latency (unit of analysis, variance reduction), supervised learning, activation functions, transformers, transfer learning, F1 versus ROC-AUC, correlated features and overfitting. The interviews were reported as conceptual and sometimes had no coding, even when a recruiter had suggested otherwise.
The Data Scientist candidate reports a conversation on stochastic processes and time-series research, a system design-style discussion on data matching across systems, and Python and SQL questions at an easy LeetCode level. The senior candidate also reports a lighter coding component and talks with a data analyst and a data engineer.
The Data Scientist candidate describes fairly informal final conversations with the hiring manager and cross-functional or leadership interviewers. The senior candidate describes a final case study in which they could present a past project or choose a provided prompt. Candidates may want to practice presenting work concisely and tying it to the team's priorities.
