Pinterest·Data Scientist·Technical Phone Screen·Intermediate

IntermediatePrefer not to say

Apr 2026Remote

Summary

Pinterest data science interview with a stats-heavy coding question. The whole thing was basically one scenario dressed up as a practical problem, which I actually appreciated more than the usual vague 'tell me about a project' stuff.

Questions Asked (1)

You have a single array of historical order values from an e-commerce platform. Write efficient Python code to compute a 95% bootstrap confidence interval for the mean using 10,000 resamples, and explain any performance choices you made.

A/B Testing & ExperimentationAlgorithms & Data StructuresTechnical Trade-offs

Author's notes

I knew bootstrapping conceptually but had never actually written it from scratch under pressure.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the problem and assumptions, then outline a vectorized NumPy implementation that resamples indices with replacement and computes means efficiently. Explain the performance rationale behind vectorization, memory considerations, and the statistical validity of the bootstrap.

Pro tip: Mention that for very large arrays, you can use a chunked or online approach to avoid memory blowup, and that setting a random seed ensures reproducibility—both are critical in production experimentation at scale.

1. Clarify the problem and assumptions

Confirm the input is a 1D array of order values, discuss whether to use the standard percentile bootstrap, and note any assumptions like independence and identical distribution.

2. Outline the algorithm

Describe the bootstrap procedure: for each of 10,000 iterations, sample n indices with replacement, compute the mean, and store the results.

3. Write efficient Python code

Use NumPy to vectorize the resampling: generate a 2D array of random indices (10,000 x n) and compute means along the appropriate axis, avoiding Python loops.

4. Explain performance choices

Justify vectorization for speed, discuss memory trade-offs (e.g., generating all indices at once vs. chunking), and mention using np.random.default_rng for modern random number generation.

5. Compute and interpret the confidence interval

Calculate the 2.5th and 97.5th percentiles of the bootstrap means to form the 95% CI, and briefly interpret it in the context of e-commerce order values.

Key Points to Mention

Vectorization with NumPy for speed and efficiency
Memory considerations and chunking for large arrays
Using np.random.default_rng for reproducible and fast random sampling
The percentile method for bootstrap confidence intervals
Statistical assumptions: independence and representative sample
Trade-offs between computational resources and accuracy

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.