okay so i just finished my databricks new grad loop. graduated december 2025 and got a referral through a friend. going to write everything i actually did to prep because the posts i found were either from 2022 or way too vague.
what the loop looks like for new grad / L3: online assessment: 2 LeetCode problems, 90 minutes. i got one easy and one medium. the medium was a sliding window variant. the OA is HackerRank-based. phone screen: 45 min with an engineer. one coding problem (medium), small talk about your projects. virtual onsite: 4 rounds back to back over one day. coding x2, systems design x1 (simplified for new grad, they don't expect distributed systems mastery), behavioral x1.
what i actually did to prep:
Spent 6 weeks on LeetCode, mostly arrays/strings/trees/graphs. The Databricks problems lean toward clean graph traversal and dynamic programming, at least from what I saw and from what others have posted. Didn't do a ton of hard problems, mostly mediums to fluency.
For systems design I did the Alex Xu book plus YouTube. For new grad the bar is "can you reason about trade-offs" not "can you design Kafka." I talked through a URL shortener and a basic messaging queue in practice and that felt like the right altitude.
Behavioral was the surprise for me. They take it seriously even for new grads. I was asked about a time I disagreed with a teammate and how I handled ambiguity in a project. Have 3-4 real STAR stories ready, actual stories not hypotheticals.
result: got through to offer stage. the comp for new grad SWE in 2026 seems to be around 150-165k base in bay area. my offer was 160k + equity. total comp ends up meaningfully higher once you factor in the RSU vest schedule.
one thing no one told me: the behavioral interviewers at databricks are really good listeners. don't rush. they'll sit in silence after you finish to see if you keep talking. let the silence breathe.