Table of contents

• Where do I apply for a Databricks interview?
• What topics should I study for?
• What are some example Databricks technical questions?
• What are some example Databricks behavioral questions?
• How long is the entire Databricks interview process?
• What is the Databricks interview like for a full time software engineering role?
• What are the levels and salary for a software engineer at Databricks?
• How should I prepare?

Where do I apply for a Databricks interview?

Apply through the Databricks careers site, or use this LinkedIn strategy to find a recruiter and message them yourself.

Databricks has a reputation for being hard to get into. Candidates describe the coding questions as medium to hard, and the loop includes a round on multithreading that many candidates have never practiced. Knowing that before you apply gives you time to prepare for it.

What topics should I study for?

The questions often feel "applied" instead of like textbook LeetCode. You'll build a small class or tool with rules to follow, and the interviewer keeps adding requirements. Study these:

• Dictionaries
• Sets (one candidate had to build a custom set class from scratch)
• Binary search
• Graph traversal
  • Graph BFS
  • Graph DFS
  • Graphs as a data structure
• Sorting
• Dynamic programming
• Deques and the sliding window idea, for "count events in the last five minutes" questions
• Heaps, for shortest path problems
• Strings and arrays, especially parsing
• Concurrency: threads, locks, condition variables, and producer/consumer queues
• System design, ideally with some feel for data systems

Most prep plans leave out concurrency, so don't let yours. If you've only ever written single-threaded Python, spend real time here. Write a thread-safe queue. Then write a version of a hit counter that many threads can call at once, and explain where a race condition could happen if you removed the lock.

What are some example Databricks technical questions?

• Design a hit counter that returns the number of hits in the past five minutes.
• Implement tic-tac-toe on a board of any size, with a fast check for the winner.
• Given a starting IP address and a count, cover the range with the fewest CIDR blocks.
• Implement a set that supports iterating over a snapshot while other code keeps changing it.
• Implement string-to-integer conversion (atoi) with all the edge cases.
• House robber: pick houses to rob for the most money without robbing two neighbors.
• Find the cheapest path through a weighted graph.
• Build a logger that many threads can write to, with messages going through a shared queue.
• Design a service that finds customers the cheapest copy of a book.
• Design a pipeline that ingests streaming data and transforms it.

Expect the question to grow while you're working on it. One candidate said the interviewers "kept making things more complicated until I got stuck or time ran out." Getting stuck at the end doesn't mean you failed. That's how the round is built. Keep talking about trade-offs as the requirements pile up.

What are some example Databricks behavioral questions?

• Tell me about a decision you made with data. What would have changed your mind?
• Tell me about a time you moved fast and it went wrong. What did you do next?
• Tell me about a time you raised the quality bar on your team.
• Tell me about a time you put the company's goals ahead of your team's.
• Why Databricks?

Databricks lists six culture principles on its careers site: customer obsessed, raise the bar, truth seeking, first principles, bias for action, and company first. Have a story ready for each one. For "why Databricks," it helps to know what the product does. Spend an hour with Spark, Delta Lake, or a free Databricks workspace so you can say something real about it.

How long is the entire Databricks interview process?

Plan on anywhere from two to eight weeks. The onsite isn't the last step. After it, Databricks checks references and a hiring committee reviews your packet, and candidates say a VP of Engineering signs off after that. References reportedly count for a lot, and they may ask for a manager plus two senior teammates. Line those people up early so you're not scrambling after a good onsite.

What is the Databricks interview like for a full time software engineering role?

Here's the most commonly reported path. The order and exact rounds vary by team and office, so check with your recruiter.

Recruiter call (about 30 minutes)

Your background, the kind of work you want, and the basics of the role.

Technical phone screen (about 60 minutes)

Live coding in CoderPad. It's usually one algorithm or data structure problem with follow-ups.

Hiring manager call (about 60 minutes)

Mostly behavioral, plus a look at your past projects. Some candidates get the behavioral part at a different point in the process.

Onsite (usually 4-5 rounds, about an hour each, often virtual and sometimes split over two days)

• Coding: algorithms. It's a medium or hard problem, and the follow-ups get harder.
• Coding: practical. Here they care more about readable code and clear thinking than the best possible runtime.
• Concurrency or systems programming. You'll write multithreaded code or reason about it, and it often mixes in operating system basics and a bit of design. Most candidates call this the hardest round.
• System design. It's often done in a Google Doc instead of a whiteboard tool, so practice writing your design out in plain text and simple lists. Questions tend to come from data platform problems like pipelines, query engines, or storage.
• Behavioral. It's sometimes run by a director or hiring manager, and it's built around the culture principles.

A few candidates have mentioned a take-home assignment of around five hours, but that's rare.

What are the levels and salary for a software engineer at Databricks?

*Note that compensation varies based on location. These estimates are likely based on the USA HCOL market.

A lot of that number is stock. Levels.fyi shows Databricks' main schedule as front-loaded: 40% of the grant vests in year one, then 30%, 20%, and 10%. That makes year one look bigger than the years after it. When you compare offers, look at year three and four pay too, not only the first year.

How should I prepare?

Get comfortable with medium and hard problems from the patterns above, and practice sticking with a problem as it changes. After you solve something, add a new rule and keep going. Give concurrency its own block of study time, and write real threaded code in Python or Java instead of only reading about it. For system design, write a few practice designs in a plain doc, since that's likely the tool you'll use. If you want a starting plan, see how to prepare for a technical interview.