Databricks Interview Questions — Page 6

Databricks·Software Engineer·Onsite - System Design / Architecture

Apr 2026

Databricks system design round centered on a bookstore price aggregation service. The deep-dive went pretty far into failure handling and consistency guarantees, which I wasn't fully prepared for. Got through the main design but ran out of time on some details, so now just waiting.

Design a bookstore price aggregation system where customers submit an ISBN, a max bid price, and a payment method. The system fans out to hundreds of partner bookstores, finds the lowest price, and either places the order automatically if it's within budget, returns the lowest price if not, or notifies the customer if the book is out of stock.
Some downstream bookstores will time out or fail. How do you aggregate partial results, and do you use a circuit breaker here?
How would you reduce the volume of calls to downstream bookstores, what caching strategy would you use, and how do you set TTL for something like book prices?
Under high concurrent traffic for a popular ISBN, how do you prevent a thundering herd from hammering downstream bookstores all at once?
Placing an order and charging the customer are two separate external calls. Which do you do first, and how do you handle a crash between the two steps? How do you prevent double-charging?

“The core design wasn't too bad.”

View Post

Databricks·Software Engineer·Onsite - System Design / Architecture

Apr 2026

Databricks system design round focused entirely on building out a small product feature end-to-end: schema design, REST API surface, and an async job pipeline. Dense interview, felt like three separate questions crammed into one session.

Design the data model and table schema for a small product feature, including entity relationships, primary keys, and indexes.
Design the REST API endpoints for the feature, covering GET, POST, PUT, and DELETE operations, including validation logic.
How would you design a long-running async job that processes work off the main request path, including the queue, worker behavior, retries, idempotency, and how clients track job status?

“I jumped straight into tables without talking through the entities first and the interviewer had to pull me back.”

View Post

Databricks·Software Engineer·Technical Phone Screen

May 2026

Databricks coding interview that started with a classic networking/IP filtering problem and then got significantly harder with a follow-up that required careful interval merging logic. The base question felt manageable but the extension exposed some real gaps in how I think about edge cases.

Given a list of CIDR rules each labeled as approve or reject, and a single query IP address, determine whether that IP should be approved or rejected using longest-prefix-match. Default to rejected if no rule matches.
Now the query is a CIDR block instead of a single IP. The block is approved only if every single IP it covers passes all relevant rules. How do you handle this?

“This part went fine.”

View Post

Databricks·Software Engineer·Technical Phone Screen

May 2026

Databricks SWE interview with a pretty gnarly tree problem that you absolutely cannot brute force your way through. The recursive structure is the whole point and if you miss that you're cooked.

A Fibonacci tree T(k) is defined recursively where T(1) and T(2) are single nodes, and for k >= 3 the root has T(k-1) as its left subtree and T(k-2) as its right. Nodes are labeled 1-based in preorder. Given k up to 60 and two node labels a and b, return the sequence of node labels along the path from a to b.

“k up to 60 was the thing that should've told me immediately: you can't build this tree.”

View Post

Databricks·Software Engineer·Technical Phone Screen

May 2026

Databricks SWE interview with a system design coding problem that looked simple on the surface but had enough moving parts to keep you busy for the full session. The core challenge was building a key-value store with sliding window hit tracking, and the design decisions around expiry strategy were where things got interesting.

Design and implement an in-memory key-value store that supports put, get, delete, and a getHits operation that returns the number of get calls on a key within a sliding 5-minute window ending at a given timestamp.

“I started with the hash map part no problem, but the sliding window hit tracking is where I got tangled.”

View Post

Databricks·Software Engineer·Technical Phone Screen

May 2026

Databricks SWE interview that was heavier on system design thinking than I expected for what felt like a coding round. The main problem sounds simple but the follow-up discussion is where they actually evaluate you.

Design a key-value store that also tracks request load, exposing a method to return the average number of operations per unit time over a rolling 5-minute window. Walk through the data structures, expiry logic, accuracy vs memory tradeoffs, and thread-safety considerations, then sketch out the class with method signatures and complexity analysis.

“I started with a deque of timestamps and evicting anything older than 300 seconds on each call, which works but I fumbled when they pushed on memory under high throughput.”

View Post

Databricks·Software Engineer·Technical Phone Screen

May 2026

Databricks software engineering interview with a pretty meaty design question around lazy evaluation. The kind of problem that sounds clean on paper but has a lot of subtle correctness issues once you start thinking about memoization and chaining.

Design and implement a Lazy Array data structure where operations like map, filter, and transform are deferred until an element is actually accessed by index. Support chaining, memoize each computed element, and avoid doing any work for elements that are never accessed.

“I started with a naive pipeline approach where each transformation just wraps the previous one as a callable, which is fine for basic laziness.”

View Post

Databricks·Software Engineer·Technical Phone Screen

May 2026

Databricks technical phone screen, graph/pathfinding problem that started reasonable and then grew legs. The state space expansion part is where things got interesting.

Given a grid representing a city with one or more transit modes, find the shortest-time or lowest-cost path from a source to a destination. Then extend the solution to penalize direction changes at intersections, where turning costs extra compared to going straight.

“I got the basic Dijkstra part fine, that was muscle memory.”

View Post

Databricks·Software Engineer·Onsite - Behavioral / Leadership

May 2026

Behavioral round at Databricks for a software engineering role. Two main areas: conflict with a teammate or stakeholder, and a deep dive into your proudest project. The project deep dive was thorough enough that it felt like a mini system design conversation layered on top of a behavioral one.

Tell me about a conflict you had with a teammate or stakeholder. What was the situation, what did you do, and how did it turn out?
Walk me through the project you're most proud of. Be ready to go deep on your specific role, the technical decisions you made, trade-offs, how you measured success, cross-functional collaboration, timelines, risks, failures, and what you'd change.

“I went with a disagreement I had with a PM over scope creep mid-sprint.”

View Post

Databricks·Software Engineer·Technical Phone Screen

May 2026

Databricks technical phone screen with a networking/IP routing problem that looked like a simple lookup question until you realized they wanted a real trie implementation. Not a lot of context given upfront, had to ask clarifying questions to understand the full scope.

Given a list of CIDR rules each tagged as approve or reject, and a query IP address or CIDR block, determine whether the query should be approved or rejected using longest-prefix-match semantics. Default to rejected if no rule matches.

“My first instinct was to just iterate through all the rules and check containment, which works but they pushed back immediately on efficiency.”

View Post