
Candidates most often describe Citi's Software Engineer process as opening with a Karat screen of Java fundamentals, debugging and short implementation tasks, with some reporting later technical, panel or managerial conversations.
$110K
Avg. Base Comp
$129K
Avg. Total Comp
Not reported
Typical Rounds
Not reported
Process Length
Most Citi Software Engineer reports here begin with a Karat screen, and it is the stage candidates describe most often. One candidate described it as a one-hour session, and several said it felt closer to working in an existing codebase than to a standalone algorithm test. Sessions typically open with Java theory: equals() and hashCode(), the difference between equals() and ==, interfaces versus abstract classes, final types, exceptions, dependency injection and JVM heap generations. They then move to reading existing code, finding a bug and implementing or improving a few methods. One candidate had to explain the approach before writing any code. Reported tasks include a restaurant order-delivery system, a toll booth journey-counting problem, string and float formatting fixes in failing test cases, and a failing unit test in a Spring Boot service that traced back to a concurrency bug.
Outcomes are mixed. Several candidates were rejected after the Karat round, one by email about 40 minutes after finishing. A few describe later conversations. One was a panel focused on Java streams and multithreading, including filtering an ArrayList of names with streams. Another was a VP-level discussion of Java backend work, Hibernate's N+1 problem, garbage collection, HashMap internals and AngularJS security. One candidate who received an offer described a process of a couple of weeks, with rounds centered on system design plus leadership, managerial and teamwork questions. Another recalled designing a payment service that handled network failures and idempotency. Round counts and timing are not consistently reported, so expect variation by team.
A sensible plan is to practice timed debugging of unfamiliar Java code with tests, rehearse a short explanation of your approach before you start coding, and review Java concurrency, collections and JVM basics. If you advance, have a few concrete stories ready about past projects, teamwork and design decisions.
Synthesized from 66 candidate reports by our editorial team.
Had an interview recently?
Share your experience. Unlock the full guide.
Real interview reports from people who went through the Citi process.
The first stage is a Karat screen: a virtual session in a platform-provided IDE, with an invigilator monitoring the whole hour. I'd heard going in that Karat handles screening for developer roles, so the format wasn't a surprise, but nobody really explained what to expect beforehand, and I got little guidance from the recruiter.
The coding portion was where I lost time. The problem statement was long, and I spent too much of the hour just reading it. Once I understood it, the solution itself was easy, which made the reading time more frustrating. The editor gives you no autocomplete or intellisense, so a small syntax error can eat several minutes, and the prompt included multiple requirements that had to be completed within a short window. I wrote the solution in Python even though the role was listed as Java, and I passed this round.
Second round. The second round was a VO with the team lead. I forgot some easy questions in that conversation, and that's what cost me.
I didn't get an offer.
Prep tip from this candidate
Practice writing working code with no autocomplete, since syntax slips in Karat's editor can burn minutes you need for the required tasks. Also practice skimming long problem statements quickly, and review basic concepts so you don't blank on easy questions in the team lead round.
Share your own interview experience to unlock all reports, or subscribe for full access.
Sourced from candidate reports and verified by our team.
Topics based on recent interview experiences.
Featured question at Citi
Write a query to show the number of users, transactions, and total order amount per month in 2020
| Question | |
|---|---|
| Size of Joins | |
| Real-Time Transaction Streaming | |
| Singly Linked List | |
| International e-Commerce Warehouse | |
| Attribution Rules | |
| Loan Model | |
| Hurdles In Data Projects | |
| Student Tests | |
| Dijkstra implementation | |
| Replace Words with Stems | |
| Measuring Text Difficulty | |
| Above Average Product Prices | |
| Concentric Circles | |
| Offer Matching API Design | |
| Concurrent LLM Serving | |
| Text Editor With OOP | |
| Seller Type Modeling | |
| Client Solution Pushback | |
| Top Daily Depositor | |
| String Palindromes | |
| Why Do You Want to Work With Us | |
| Top Charting Song | |
| Your Strengths and Weaknesses | |
| Index Fund Return | |
| Unified Event Pipeline | |
| 2nd Highest Salary | |
| Employee Salaries | |
| Closest SAT Scores | |
| Merge Sorted Lists |
Synthesized from candidate reports. Individual experiences may vary.
One candidate was contacted by a recruiter after a friend's referral, and another received an HR call to schedule a 30-minute technical phone interview. Confirm the format of the first technical round when you schedule it.
Most reports here start with a virtual Karat interview, which one candidate described as one hour long. It typically opens with Java fundamentals such as equals() and hashCode(), interfaces versus abstract classes, final types, exceptions, dependency injection and JVM heap generations, then moves to a hands-on coding block.
The main reported task is reading existing code, finding and fixing a bug, then implementing or improving a few methods. Examples include a restaurant order-delivery system, a toll booth journey-counting problem and a failing unit test in a Spring Boot service. One candidate had to explain the approach before coding.
A few candidates describe follow-up rounds with a panel or a VP-level interviewer. Topics included Java streams, multithreading, Hibernate's N+1 problem, garbage collection, AngularJS security, system design and past experience. Formats differ between reports, so expect variation by team.
One candidate who received an offer described leadership, managerial and teamwork questions alongside system design. Outcomes range from rejection shortly after the Karat screen to an offer.