Databricks · Software engineer · IC SWE (L3-L5)
Onsite Coding (algorithms + data-processing twist)
Onsite algorithms round over Google Meet in a runnable editor
What this round is judged on
- Correctness · code that runs and passes tests, not just a sketch
- Algorithmic depth · graphs, heaps, trees, tries, some DP
- Production quality · clean, readable, edge-case-covered code
- Complexity & tradeoffs · Big-O and how the solution scales
Where this round sits
Databricks’s published process. Lime marks the stage this round runs.
- 01Recruiter / Talent Acquisition Screen
- 02Online Assessment (OA)
- 03Technical Phone Screen
- 04Virtual Onsite ('Loop')this round
- 05Reference Checks
- 06Hiring Committee + VP of Engineering sign-off
What happens in each stage›
- Recruiter / Talent Acquisition Screen
- ~30 min, non-technical. Background, motivation, and target team/level fit. Confirmed by Databricks' own 'Connecting with Talent Acquisition' stage.
- Online Assessment (OA)
- ~70-min proctored coding test (CodeSignal/HackerRank), typically ~4 questions mixing standard data-structure work with a data-manipulation twist. Common for new-grad/many IC pipelines; some experienced candidates skip straight to the phone screen. Async, so simulated by the phone-screen coding round. Corresponds to Databricks' 'Skill assessments' stage.
- Technical Phone Screen
- ~60 min live coding in a shared runnable editor (CoderPad-style). Medium-to-hard algorithmic problems; graph traversal recurs. Code is expected to compile, run, and pass tests.
- Virtual Onsite ('Loop')
- 5-6 interviews, fully virtual over Google Meet (confirmed on Databricks' own prep page). Stable core: 2 coding rounds (one with a data-processing slant), 1 dedicated concurrency/multithreading round, 1 system design, 1 behavioral/values with a hiring manager.
- Reference Checks
- Explicitly listed as stage 6 of Databricks' published 7-stage process.
- Hiring Committee + VP of Engineering sign-off
- Decision made by committee from the full packet rather than round-by-round, which is why turnaround feels slow. Committee/VP specifics are candidate-reported, not officially published.
Sources
Built from what Databricks publishes about its own process and from real interview data. No leaked question lists. The questions you get are generated against your own resume, so they are not from anyone else's interview.
7-stage process, Google Meet as the interview tool, exact behavioral question stems, and an engineering-specific prep guide.
Culture language ('innovators, builders and truthseekers'), early-career/engineering hiring context. Note: does NOT publish a verbatim six-value list.
Factual company description: unified data-and-AI lakehouse platform; creators of Apache Spark, Delta Lake, MLflow, Unity Catalog.
Confirms role-specific prep guides exist. Adjacent role (field engineering), not SWE, so used only as corroboration that prep material is published.
Leveling and compensation context only; comp figures are informational and single-sourced. Not used for round mechanics.
Sources last checked . Hiring loops change, so this date is the honest limit on everything above.
Other Databricks rounds
- Recruiter / Talent Acquisition Screen
Conversational and non-technical
- Technical Phone Screen (live coding)
One hour of live coding in a shared runnable editor
- Concurrency / Multithreading (signature round)
The defining Databricks round: ~1 hour writing correct multithreaded code while the interviewer actively watches for the race condition you didn't catch
- System Design (distributed systems)
~1 hour of spoken design, often in a shared Google Doc
- Behavioral / Values (Hiring Manager)
45-60 min with a hiring manager