Coding, Data Structures, And Parsing

What's being tested
Candidates must show they can parse messy input reliably and choose data structures and algorithms that meet correctness, time, and memory constraints at scale. Interviewers look for clear problem framing, edge-case handling, streaming vs batch trade-offs, and when to use approximations.

Core knowledge

Worked example
Framing: "Top-k active users per minute from newline-delimited logs" — first ask input size, memory limits, timestamp format, and whether approximate results are allowed. Decide streaming processing if input > memory: parse each line to extract user ID and minute bucket, maintain hashmap minute→(user→count) and a min-heap per minute for top-k, or use Count-Min plus a heap for heavy hitters to reduce memory. Discuss time-window retention, handling late or out-of-order timestamps, and fall back to external sort or map-reduce for exact global aggregation if streaming state is too large.

A common pitfall
A tempting but wrong approach is to read all data into in-memory arrays and then parse/sort blindly; this misses scale and edge cases like quoted CSV fields or malformed JSON. Another common mistake is using regex to parse complex nested formats, which fails on escaped characters and embedded delimiters. Also watch for not asking about input size or error handling — those constraints change the whole design.

Further reading