Oscar Insurance software engineering one candidate reports paired-programming coding, a later coding interview, and a 60-minute workflow-focused system design discussion.
$173K
Avg. Base Comp
$206K
Avg. Total Comp
3
Typical Rounds
Not reported
Process Length
The reported Oscar Insurance Software Engineer process begins with a paired-programming, LeetCode-style coding round. One candidate characterized that question as medium-plus difficulty and centered it on recursion or dynamic programming. Prepare to explain your approach as you work: clarify inputs, walk through the recurrence or state choices, and communicate trade-offs rather than treating the exercise as a silent solve.
The same candidate then described a later stage containing two interviews: another coding round, with a problem similar to LeetCode’s “Heaters,” and a 60-minute system design discussion. The most specific design signal is a campaign builder for human agents. The task involved configurable, branching workflows with no fixed single next step, terminal outcomes, drafts, immutable finalized campaigns, database persistence, and running campaigns for target sets.
For that discussion, practice turning an ambiguous workflow into a clear model: identify the campaign, stages, transitions, terminal states, version lifecycle, and execution state separately. Narrate assumptions and offer alternatives when requirements are underspecified. The available report is from one candidate, so it cannot establish how often each topic appears.
Synthesized from 1 candidate report by our editorial team.
Had an interview recently?
Share your experience. Unlock the full guide.
Real interview reports from people who went through the Oscar Insurance process.
First round was a paired-programming LeetCode-style round. I would place the question at a medium+ level with recursion/DP.
I moved to a second round with two interviews. One was a coding round with a question very similar to the "Heaters" problem on LeetCode.
The part I failed was the system design round, where the interviewer did not provide yes-or-no answers to stated assumptions. The question was to build a "campaign builder" that human agents could use to design and launch campaigns. The interview lasted 60 minutes.
A campaign is a defined workflow with no deterministic previous or next step. It has configurable stages, each with an undetermined number of next steps, and a terminal state of "campaign concluded" for any reason.
The design needed to support designing and saving drafts and final immutable campaigns in a database, then running campaigns for target sets. The example workflow used audience selection, language branching, phone calls, email/text follow-ups, waits, response handling, and successful or unsuccessful terminal states.
Prep tip from this candidate
Practice medium-to-hard recursive or dynamic-programming coding problems, then rehearse a system design for a branching workflow: model stages, transitions, terminal states, draft versus immutable versions, persistence, and execution for a target set. State assumptions clearly and explain how your design handles ambiguous requirements.
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 Oscar Insurance
Explain what a p-value is to someone who is not technical
| Question | |
|---|---|
| Confidence Interval Explanation | |
| Why Do You Want to Work With Us | |
| 2nd Highest Salary | |
| Employee Salaries | |
| Prime to N | |
| Top 5 Turnover Risk | |
| Bagging vs Boosting | |
| Size of Joins | |
| Rectangle Overlap | |
| Hurdles In Data Projects | |
| Total Spent on Products | |
| The Brackets Problem | |
| Google Maps Improvement | |
| Valid Anagram | |
| Type-ahead Search | |
| Always Excited Users | |
| Sort Strings | |
| String Mapping | |
| Three Zebras | |
| Precision and Recall | |
| Find Duplicate Numbers in a List | |
| Target Indices | |
| RAG Strict Source Control | |
| Transformer Encoder Layer | |
| Prime Numbers Identification | |
| Duplicate Rows | |
| Implementing the Fibonacci Sequence in Three Different Methods | |
| Type I and II Errors | |
| Digitizing Student Test Scores |
Synthesized from candidate reports. Individual experiences may vary.
A candidate reports a paired-programming LeetCode-style round with a medium-plus recursion or dynamic-programming problem. Practice explaining the state definition, base cases, complexity, and implementation decisions while coding collaboratively.
The candidate reports a second coding interview in a later stage, using a question similar to LeetCode’s “Heaters.” Candidates may benefit from reviewing distance, sorting, and boundary-case reasoning alongside a clear verbal walkthrough.
The candidate reports a 60-minute design prompt for a campaign builder used by human agents. Prepare to model branching stages, transitions, terminal states, draft and immutable campaign versions, database storage, and execution for target sets; requirements may need to be clarified through assumptions.
