Upstart·Data Scientist·Technical Phone Screen·Intermediate

Intermediate

Jun 2026

Summary

Upstart data scientist interview with a Python coding question built around a realistic API response scenario. Pretty straightforward if you know your way around groupby or defaultdict, but the 'sorted by descending average' part is easy to forget under pressure.

Questions Asked (1)

Given a list of user event objects (each with a user_id, event_type, value, and timestamp), write Python code that returns a dictionary mapping each user_id to the average value of their events, sorted by descending average value.

Algorithms & Data StructuresAPI & IntegrationsProduct Analytics & Metrics

Author's notes

I went with defaultdict to accumulate sums and counts in one pass, then divided at the end and sorted with sorted() using a lambda on the value.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, clarify the requirements: what to do with missing values, whether to sort the dictionary itself or return a sorted list of tuples, and the expected output format. Then, implement an efficient solution using a dictionary to accumulate sums and counts per user, compute averages, and sort by descending average. Finally, discuss time and space complexity and potential edge cases.

Pro tip: Mention that Python dictionaries preserve insertion order (since 3.7), so if you need a sorted dictionary, you can build it from sorted items; otherwise, returning a list of (user_id, average) tuples sorted by average is often more practical. Also, consider using collections.defaultdict for cleaner accumulation.

1. Clarify requirements and edge cases

Ask about handling missing values, non-numeric values, empty input, and whether the output should be a dictionary or a sorted list. Confirm if sorting is by average value descending and if ties need a secondary sort.

2. Choose data structures

Use a dictionary to map user_id to a list of values, or two dictionaries for sum and count. Consider collections.defaultdict for efficient accumulation.

3. Compute averages

Iterate through events, accumulate sum and count per user, then compute average for each user. Handle division by zero if a user has no events (though unlikely).

4. Sort by descending average

Sort the user averages using sorted() with key=lambda x: x[1], reverse=True. If returning a dictionary, build it from the sorted items.

5. Analyze complexity and test

State time complexity O(n + m log m) where n is number of events and m is number of users, and space O(m). Walk through a small example to verify correctness.

Key Points to Mention

Use of dictionaries for O(1) average-case lookups and updates.
Handling of edge cases: empty list, users with no events, non-numeric values.
Sorting stability and tie-breaking (e.g., sort by user_id if averages are equal).
Time and space complexity analysis.
Python-specific features: defaultdict, sorted() with key, dictionary insertion order.
Potential for using pandas or numpy for larger datasets, but pure Python is sufficient here.

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