314,552 interview questions from 6,000+ companies.
Tests influence without authority through data-driven marketing analysis, stakeholder alignment, and ownership of a measurable business outcome.
Tests ownership under ambiguity, prioritization, and communication during an unclear production problem.
Explain how a primary metric differs from a guardrail metric and how both are used in A/B test decisions.
Pick a North Star Metric that reflects customer value, business impact, and long-term product health.
Tests communication of complex data to non-technical stakeholders, including clarity, stakeholder management, and actionable storytelling.
Determine sample size and power for a customer survey or experiment, including MDE, guardrails, and a disciplined decision rule.
Tests how a candidate makes a quality-vs-speed trade-off, communicates risk, and owns the outcome.
Investigate sample ratio mismatch and decide whether an experiment readout is trustworthy enough to ship.
How to evaluate a production model using calibration, thresholds, and confusion matrix tradeoffs.
Tests influence without authority by using data to challenge leadership assumptions and drive an operational decision.
Assess how CUPED changes variance and sample size in an online experiment, and decide whether the uplift is detectable.
Tests whether you can communicate statistical thinking clearly, own the analysis end-to-end, and adapt your message to different audiences.
Explain when network interference threatens an A/B test, how it biases estimates, and how to redesign the experiment safely.
Design a browse-surface A/B test that measures true lift while guarding against short-term novelty effects and premature shipping.
Tests ownership and communication in ambiguous data-cleaning work, especially how you used Python/Pandas to turn unreliable data into a trusted output.
Define how to measure whether an AI feature improves engagement, repeat usage, and downstream product value.
Pick a primary success metric and guardrails for a new feature, balancing user value with broader product health.
Design an experiment analysis that uses CUPED to reduce variance and detect a small lift with enough power.
