[Second round interview]

60-minute live coding session on HackerRank

One engineer observing and interacting with you throughout

You get a file-like string input that simulates a messy CSV (not just an array or list)

First, you need to write code to robustly parse and clean the data

You'll clarify weird formatting and ask questions as needed—the interviewer will give details if you ask

Must process the data as a stream, not reading the whole thing into a DataFrame by default

After parsing, you’ll answer a set of concrete business queries based on the cleaned data

At the end, there are a few minutes for your own questions

Interview Questions how many employees have a salary strictly over 30,000.Identify which employee has held their job the longest by using start and end dates, treating missing end dates as “today” and missing start dates as “skip”.Find the employee with the second-highest salary. You may need to decide how to handle missing names by using a recommended placeholder like “(unknown)” so the row isn’t dropped for the other queries. Clarify how to deal with rows that are malformed, have extra or missing fields, or contain blank lines. Explain your data structures and approach as you go, especially concerning performance and edge cases. You have the option to use native Python types and modules, as pandas is probably not available. You don’t have to provide a tie-breaking rule unless asked, but if you do, picking the lexicographically smallest name is reasonable. Overall, strong candidates can parse and clean up messy, inconsistent tabular input, make and explain reasonable real-world assumptions, use Python constructs like csv.reader, datetime.fromisoformat, and simple lists/dicts, and communicate their approach and edge-case handling out loud.