Ethos 前端面试指南

For Ethos, the life-insurance company, practice debugging an existing React application and explaining the behavior of a feature across the client and server. A September 2026 senior full-stack candidate encountered React board debugging, an interval problem, and full-stack system design.

Frontend applicants should confirm the backend scope of their own loop. Alongside the interview exercises, practice conditional application forms and interrupted user journeys: both feature in Ethos's engineering articles about its insurance application.

Interview process

The September candidate described backend coding, frontend debugging, system design, and a hiring-manager deep dive into prior work. Their design discussion covered both browser behavior and backend reliability. The round details below come from that senior full-stack interview.

Ask whether you will receive an existing project, which language is expected for general coding, and whether the design round covers storage and service failures. Confirm IDE, documentation, and AI permissions before rehearsing in the intended environment.

Coding rounds

Debugging an existing React board

The frontend exercise involved a buggy Trello/Jira-style board, including lane categorization, filtering, sorting, and routing. Practice reading another developer's components and tracing how a record becomes a visible card. Identify whether a failure comes from the underlying data, a derived view, or navigation state before changing the code.

Use stable record IDs and check that a filter does not accidentally overwrite the complete dataset. When sorting, consider whether an in-place mutation changes state used elsewhere. After fixing an issue, test combined operations: filter the board, change a card's lane, clear the filter, and navigate away and back.

Several smaller exercises cover the state-handling problems in that board. Filtering, selection, and edits all share the same local data in Users Database. Once the basic CRUD behavior works, add status-based grouping and sorting as follow-ups.

To isolate the transformation logic, work through Data Selection, which combines filters and optional merging without changing the input objects. Its explicit ordering rules make it useful for checking whether a derived view accidentally loses or rearranges records. For the board, keep lane grouping separate from the underlying records and connect the active view to the URL.

An edit to one card should also leave unrelated state intact. In Immutability Utilities, only objects along the updated path are copied; untouched branches keep their references. Model cards in an object keyed by ID to practice this behavior, since the base exercise does not support array updates.

Interval and date problems

The backend coding task used records containing a medication name and start/end dates, asking for the maximum number of overlapping intervals on a day. Prepare the interval-counting algorithm and clarify whether both endpoints count before implementing it.

The same maximum-overlap calculation appears in Minimum Meeting Rooms Needed, where a room can be reused as soon as a meeting ends. If a date-range task uses inclusive end dates, the ordering of start and end events must change. Test touching intervals, identical ranges, and an empty input.

System design rounds

The September candidate designed a stock-trading platform across the frontend and backend. The discussion included database choices, latency, consistency, durability, and failover. Prepare to explain both a responsive interface and the server guarantees behind a user action.

As practice, trace an order from a button click to a confirmed result. Distinguish a pending submission from a completed transaction, and explain how the server prevents a retry from creating a duplicate operation. Then consider a disconnected client: what becomes stale, what can still be shown, and how does the client reconcile its state after reconnecting?

Design a reusable table for orders or positions, drawing on Data Table. Its base requirements cover configurable columns, custom cells, scrolling, and performance. Once that design is complete, introduce live updates and explain what happens to the visible rows and the user's current selection as data arrives.

Use the Front End System Design Playbook for the client-side structure. For a full-stack loop, continue into persistence and recovery: specify which operations require durable writes and what the user sees during a service failure.

Project and hiring-manager deep dives

The candidate's hiring-manager conversation explored previous projects, including resilience, failover, and improvements to how the team operated. Choose a project where you can follow a failure from detection through recovery. Explain your own decisions and what changed afterward in monitoring, testing, or release practices.

If your experience is primarily frontend, use a concrete example such as recovering an interrupted workflow or diagnosing failed requests. Explain the server contract you relied on and who owned the backend changes.

Product preparation

Ethos's February 2022 consumer-engineering overview describes an application that asks different questions based on earlier answers and lets customers return to an unfinished application. Its February 2023 frontend article discusses experimentation and collaboration with product managers, designers, and analysts. Use these older articles to understand the product problems and engineering decisions.

For product-inspired practice, sketch a conditional application form using the public product explanations. Decide what happens when an earlier answer invalidates a later one, how saved progress is restored, and how a user recovers from a failed submission. Keep private form values out of routine analytics events.

Recommended preparation strategy

1. Clarify the frontend/backend split: Confirm whether your interview follows a full-stack loop and whether you will modify an existing React project.
2. Debug before rebuilding: Practice reproducing a board defect and making a small fix. Use the UI coding collection for additional stateful interfaces, and check React behavior with fundamentals quiz questions.
3. Rehearse interval boundaries: Solve the overlap problem, then change its endpoint rules and update the tests. Keep calendar-date interpretation separate from the counting algorithm.
4. Trace failures through a design: Pick a scenario from the system design collection, then add a lost response and a reconnect. For full-stack preparation, explain persistence and duplicate prevention as well as client recovery.
5. Study the product and one past project: Read the engineering articles below, sketch a resumable form, and prepare a project discussion that connects a user-facing improvement to a specific reliability or team-process change.

Company blog posts

• Overview of Consumer Engineering at Ethos: A February 2022 overview of conditional application flows, returning customers, and the systems supporting the insurance journey.
• Coding for Instant Life Insurance at Ethos: A February 2023 frontend engineer's account of application development, experimentation, and work with product and design partners.