Karat·Software Engineer·Technical Phone Screen·Intermediate

IntermediatePrefer not to say

May 2026Remote

Summary

Karat screen for a software engineer role, one problem the whole time. It was a multi-table join puzzle dressed up as a coding question and I spent way too long second-guessing the matching logic instead of just writing the thing.

Questions Asked (1)

You're given three tables as 2D arrays (customers, orders, products) with no headers. Write code to produce a mapping from userId to customerName by joining them. Orders may have a null email field, in which case you need to infer the customer by matching orderTotal against quantity times unitPrice from the other tables.

Algorithms & Data StructuresData ModelingTechnical Trade-offs

Author's notes

The email path was easy enough, just a lookup.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, clarify the table schemas and the join logic, especially how to handle null emails by inferring the customer through order totals. Then, outline a step-by-step algorithm that builds lookup maps for customers and products, processes orders, and outputs the userId-to-customerName mapping. Finally, discuss edge cases and potential optimizations.

Pro tip: Explicitly state your assumptions about column order and data types, and propose a fallback strategy if the inferred matching fails (e.g., logging or defaulting). This shows attention to detail and robustness.

1. Clarify schemas and assumptions

Ask about the column order in each table (e.g., customers: [userId, name, email], orders: [orderId, userId, email, orderTotal, productId, quantity], products: [productId, unitPrice]). Confirm that orderTotal equals quantity * unitPrice when email is null.

2. Build lookup maps

Create a map from userId to customerName using the customers table. Also, create a map from productId to unitPrice using the products table.

3. Process orders and infer customers

For each order, if email is not null, find the customer by email (requires a map from email to userId or customerName). If email is null, compute expected total = quantity * unitPrice (using productId) and match it to orderTotal to find the customer.

4. Handle edge cases and ambiguities

Discuss what to do if multiple customers match the same total (e.g., pick the first, log a warning, or use additional fields). Also, consider missing productId or quantity, and invalid data.

5. Return the mapping

Collect the results into a dictionary mapping userId to customerName. Ensure the output includes all orders, even those with null emails that were successfully inferred.

Key Points to Mention

Time and space complexity: O(n + m + k) where n, m, k are sizes of customers, orders, products.
Use of hash maps for efficient lookups (email to userId, productId to unitPrice).
Handling of null emails by inferring through orderTotal matching.
Potential ambiguity when multiple customers have the same order total; discuss resolution strategies.
Data validation: ensuring orderTotal matches quantity * unitPrice within a tolerance for floating-point numbers.
Scalability: if data is large, consider streaming or database joins instead of in-memory processing.

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