Just cleared the Twilio OA last week so posting while it's fresh. Applying for a software engineer role, 2025 grad.
The online assessment: Two coding questions, 90 minutes total. Both were on HackerRank. The first one was a medium-level string/parsing problem, the kind where you have to handle edge cases carefully. Think: given a sequence of API webhook events, reconstruct the order of delivery and identify any duplicates. Very Twilio-flavored.
The second was closer to a medium-hard graph/BFS problem. Not textbook "find the shortest path" though. It had a real-world framing about message routing between nodes where some nodes have capacity limits.
Difficulty: both felt like LC medium. I didn't see anything LC hard. But the edge cases on the first problem caught me because I initially didn't handle empty strings properly.
Time distribution: I spent about 35 minutes on Q1, 45 on Q2, and the last 10 writing test cases. Submitting tests felt important because the problem statements mentioned they care about correctness over speed.
After the OA: Got a recruiter call 10 days later. They said they look at passing test cases and code quality (readability, not just correctness). So don't just smash it in, add comments.
Languages: I used Python. Saw in another post someone used Java, said it was fine. They list Python, Java, C++, JavaScript as accepted.
One thing I wasn't sure about: the OA doesn't have a proctoring webcam, at least mine didn't. Just the HackerRank environment.
If you're studying: I'd do 15-20 LC mediums with a focus on strings, graphs, and queue-based simulation. That should cover it. Good luck.