Interview guides · Brunswick

Brunswick Software Engineer Interview Guide

Brunswick Software Engineer interview guide: three reported rounds, behavioral and technical questions, a 7-day plan and answer structures for each.

The 60-second brief

Brunswick appears in candidate reports as an employer with marine products, and the Software Engineer projects described to candidates range from embedded software for marine applications to data analytics platforms that give operational insight. In this role you would write and maintain application code, turn product managers' requirements into technical work, review teammates' code and take part in Agile ceremonies such as sprint planning. Reports say little beyond that, so this guide prepares you for the question types and the languages listed (Python, Java, C++) and does not guess at product details.

This guide is for candidates interviewing for a Software Engineer position at Brunswick. It covers the three reported rounds (Phone Screening, Technical Assessments, In-Person Interviews), answer structures for the twelve reported questions (behavioral, problem-solving, coding and system design), and a seven-day plan that sequences them.

Candidates report three rounds over roughly three to five weeks: a phone screening, technical assessments, then in-person interviews with several team members. Candidates describe a supportive tone and open dialogue.

Focus areas

Behavioral Interviewing
Resume/Experience Review
Technical Interview (General)

Overview

The reported question set is a mix. About half is open-ended: describe a technical challenge, explain the software development lifecycle, handle a difficult teammate, take the lead on a project, and prioritise under several deadlines. The rest is technical: name the languages you are proficient in, speed up a slow application, react to a serious bug just before launch, design for a sudden jump in traffic, solve a Python problem, explain the time complexity of your solution, and build a data structure that manages user sessions. Preparation therefore splits into story work and technical work, and neither can be skipped.

The engineering described to candidates spans embedded software for marine products and data analytics platforms. You cannot know which team you will meet, so keep your language answers flexible: the requirements candidates describe list Python, Java or C++, with AWS or Azure and Agile experience as extras. Be ready to name the projects where you used each language you claim, and to say what the language was good and bad at for that project.

The responsibilities candidates describe include code review, working with product managers and Agile planning, which makes the lifecycle and teamwork questions practical rather than theoretical. Answer them with specific episodes from your own work: what the team did in a sprint, how a review changed a pull request, how a requirement turned into tasks.

Candidates describe a conversational atmosphere with open dialogue rather than a rigid quiz. That favours people who think aloud: narrate your assumptions, say what you would measure before changing anything, and state the complexity of your solution unprompted. The plan below alternates spoken story practice with timed coding and diagnosis practice for that reason.

The Brunswick interview process

Candidates describe three stages, in this order: Phone Screening, Technical Assessments and In-Person Interviews, over roughly three to five weeks. Reports do not break the stages into more detail than the descriptions below.

Phone Screening

Candidates report an opening phone call about your background and how you fit the role. No technical format is described for it, so expect a conversation about your resume, the languages you have used and why this role. Prepare a clear account of your last two projects, the question about which programming languages you are proficient in, and a short explanation of how your teams run the software development lifecycle. Have a few questions ready about the team and the work.

What to demonstrate

How clearly you summarise your background and experience
Whether your skills and interests match a Software Engineer role
How well you explain your own work in conversation

How to prepare

Write a resume walkthrough that names, for each recent project, the language, your role and one decision you made
For each of Python, Java and C++ that you list, prepare one project and one honest strength and weakness of that language
Prepare a short account of the lifecycle and methodology you actually worked in, using one real sprint or release as the example
Write two or three questions about the team's projects and way of working to ask at the end

Technical Assessments

Candidates report technical evaluations in which you demonstrate your skills. Reports do not say whether this is live coding, a take-home or a discussion, so prepare for all three. The reported technical questions give the types to practise: a Python coding problem with a complexity explanation, a data structure for user sessions, diagnosing a slow application, handling a serious bug before a launch, and designing for a sudden traffic increase.

What to demonstrate

Working technical skill shown through code or reasoning, not description alone
Your ability to explain the approach you chose and its complexity
How you structure an open diagnosis or design problem

How to prepare

Solve Python problems on arrays, hash maps and linked structures, stating time and space complexity before you are asked
Implement a session store with create, lookup, refresh and expiry, and be ready to compare a hash map with expiry against alternatives
Rehearse a fixed order for a slow application: reproduce, measure, find the bottleneck, change one thing, measure again
Outline a traffic-spike design out loud covering caching, load balancing, queues and degraded modes

In-Person Interviews

Candidates report several interviews with various team members, with collaboration and communication as the stated focus. Expect many conversations, so keep separate stories ready for the behavioral questions candidates report: the difficult teammate, taking the lead on a project, prioritising under several deadlines, and a technical challenge you resolved. Candidates also mention persuading a colleague and handling feedback from peers as example topics.

What to demonstrate

How you collaborate and communicate with people on different teams
How you handle disagreement, feedback and competing priorities
Whether you can tell a specific, honest story about your own contribution

How to prepare

Write four stories (difficult teammate, leading a project, competing deadlines, technical challenge) in a situation, action, result structure and rehearse them aloud
Prepare a story about persuading a colleague with evidence and one about acting on critical feedback
Vary the stories you tell across different conversations so each one adds new information
Prepare questions for each type of colleague you may meet: engineers, product managers and designers

Where candidates lose points

Answering the languages question with a list of names instead of projects

For each language you claim, prepare one project, your part in it and one honest limitation of the language. Python, Java and C++ are the languages candidates describe, so cover whichever you list.

Reciting lifecycle phases and methodology definitions with no team you worked on

Anchor the lifecycle answer in one real release or sprint, including how code review and testing happened. Keep the definitions to one sentence.

Casting the difficult teammate as the villain, or telling a story where you only complained

Describe the behaviour and its effect, the conversation you started, the concrete agreement that followed and your own part in the friction.

Jumping to fixes for the slow application or the launch bug before measuring or triaging

Use a fixed order out loud: reproduce, measure, isolate, change one thing, verify. For the launch bug, add severity, workaround or rollback, and communication to stakeholders.

Stating a complexity without deriving it, or writing a session structure with no expiry or eviction story

Name the loop or data-structure operation that drives each term of the complexity. For sessions, cover lookup, refresh, expiry and the memory growth of expired entries.

Questions & practice

Solve this coding problem using Python: [insert specific problem here].

Medium · Coding / Algorithms

Approach

Restate the problem with one small example, then pin down input size, value ranges, duplicates, empty input and the expected output format. The size tells you which complexity you need.
Say the brute-force solution and its cost out loud, then improve it by choosing the structure from the access pattern: dict or set for O(1) average lookups, heapq for top-k (a min-heap, so negate values for max), collections.deque for BFS queues, two pointers or bisect on sorted data.
Write idiomatic Python: enumerate, zip, collections.Counter and defaultdict, and ''.join for building strings. Avoid list.pop(0) in a loop (O(n) per call) and deep recursion, because CPython's default recursion limit is 1000.
Test by tracing your own example by hand, then the edge cases: empty input, a single element, all duplicates, negatives, and the largest size.
State the final time and space complexity and the one trade-off you made, for example extra O(n) memory for a hash map to drop from O(n^2) to O(n).

Follow-up

Can you reduce the extra space, and what would it cost in time?
How would the solution change if the input arrived as a stream too large for memory?
Which test cases would you add before merging this?

Explain the time complexity of your solution.

Medium · Coding / Algorithms

Approach

Count the work per input element: nested loops over the same input give O(n^2), and halving the search space each step gives O(log n). Sorting is O(n log n) because Python's Timsort has that worst case.
Name the cost of each Python operation you used: dict/set lookup O(1) average and O(n) worst case, x in list O(n), list append O(1) amortized, slicing O(k) because it copies, heappush/heappop O(log n), heapify O(n).
For recursion, write the recurrence and solve it, for example T(n) = 2T(n/2) + O(n) gives O(n log n) for merge sort. Note that memoization turns exponential naive recursion into O(number of states x work per state).
Give space complexity separately, including the recursion stack depth and any copies, and say whether you mean worst, average or amortized cost.
Tie the result to the constraint: for n around 10^5, O(n^2) is about 10^10 steps and too slow, while O(n log n) is fine.

Follow-up

What is the worst case for your hash-map solution, and when could it happen?
Why is appending to a Python list O(1) amortized rather than O(1) every time?
What is the space complexity of your recursive version, counting the call stack?

How would you implement a data structure to manage user sessions?

Medium · Coding / Algorithms

Approach

Clarify the operations: create a session at login, look one up per request, extend it on activity (sliding expiry) or not (absolute expiry), delete it on logout, and possibly log out all of a user's sessions.
Core structure: a hash map from session id to a record (user_id, created_at, last_seen, expires_at) for O(1) average lookup, plus a map from user_id to a set of session ids so revoking every session for a user is cheap.
Expiry: check expires_at lazily on every read, and sweep in the background with a min-heap keyed on expiry time (O(log n) push and pop, with lazy deletion of stale heap entries). For a capacity cap with sliding expiry, an OrderedDict with move_to_end and popitem(last=False) gives O(1) LRU eviction.
Security: generate ids with Python's secrets module (for example secrets.token_urlsafe(32)), never sequential or random.random, and rotate the id at login to prevent session fixation.
Concurrency and scale: guard the in-memory version with a lock, and for several app servers move it to a shared store such as Redis with a per-key TTL (SET with EX, refreshed on activity) so any server can validate a request.

Follow-up

How would you support "log out of all devices" efficiently?
What changes if the session store must survive a restart or run across several servers?
Would you use signed stateless tokens instead, and how would you revoke one?

Write the update path that detects a concurrent edit

Medium · optimistic concurrency · transactional outbox · audit history

resource carries version INT NOT NULL DEFAULT 1. resource_revision holds revision_id, resource_id, version, actor_user_id, change_kind, patch JSONB, request_id, created_at with UNIQUE (resource_id, version). outbox_event holds aggregate_type, aggregate_id, aggregate_version, event_type, payload, status. A PUT carries the version the client read. Write the exact statements for the single transaction that applies the edit, records the revision and enqueues 'resource.updated', and give the handler's branch on zero affected rows. Then say what PostgreSQL 16 does under READ COMMITTED when two of these updates hit one row at once.

Approach

One transaction, three writes, no network call inside it: UPDATE resource SET title = $3, version = version + 1, updated_at = now() WHERE resource_id = $1 AND tenant_id = $4 AND version = $2; then INSERT the resource_revision row at version $2 + 1; then INSERT the outbox_event row at the same aggregate_version. The event goes to a table rather than a broker because no transaction spans both.
Branch on the affected-row count before doing anything else. Zero has three causes — stale version, wrong tenant, row gone — so re-read once and map to 409 carrying the current version, or 404 for an id outside the caller's tenant, which also stops the endpoint confirming that another tenant's id exists.
State the engine behaviour instead of assuming it. Under READ COMMITTED the second UPDATE blocks on the row lock, and when the first commits PostgreSQL re-evaluates the WHERE clause against the newly committed row, so the version predicate now fails and the statement reports zero rows. Under REPEATABLE READ the identical collision raises SQLSTATE 40001 instead, so the handler must fold both shapes into one conflict response.
Keep UNIQUE (resource_id, version) even though the predicate already serialises writers. It is what makes a lost update unwritable if any other path ever reaches the revision table, and it converts a logic bug into 23505 rather than into a silently missing history row.

Follow-up

A client sends the version it read ten minutes ago and the resource has moved three versions. What is in your 409 so it can resolve the conflict without a full re-fetch?
Two editors, two disjoint fields, no overlap. Does your answer still refuse the second write, and should it?

Keep soft-deleted accounts from blocking re-registration

Easy · soft delete · unique constraints · engine differences

app_user holds user_id, tenant_id, email CITEXT, password_hash (NULL for SSO principals), email_verified_at, auth_version, status ('invited','active','suspended','deactivated'), created_at, updated_at, deleted_at. Two live accounts for one address inside a tenant must be impossible, but an address freed by a soft delete must be reusable, and the same tenant may delete and re-register it repeatedly. Write the uniqueness DDL for PostgreSQL 16, then the equivalent for MySQL 8 where partial indexes do not exist, and say what each permits once three deleted rows already hold that address.

Approach

Start from what is actually unique: not (tenant_id, email), but (tenant_id, email) among live rows. PostgreSQL says that directly — CREATE UNIQUE INDEX app_user_live_email ON app_user (tenant_id, email) WHERE deleted_at IS NULL. A full constraint over the same two columns burns the address permanently the first time someone deletes an account.
Keep case-insensitivity in the type or the index, never in the application: CITEXT as given, or UNIQUE (tenant_id, lower(email)) as an expression index where the extension is unavailable. A case-sensitive unique column is exactly how two accounts for one human appear.
For MySQL 8 the predicate has to move inside the key: add a discriminator column that is a constant 0 while the row is live and is set to user_id on delete, with UNIQUE (tenant_id, email, deleted_marker). Live rows share the constant and still collide; deleted rows differ from each other and stop colliding.
State the NULL variant and its dependency: leaving the marker NULL for deleted rows also works, because a unique index treats NULLs as distinct — true in MySQL, and true in PostgreSQL only under the default NULLS DISTINCT, which PostgreSQL 15 lets you reverse. Check the polarity against the three existing deleted rows: constant-on-live is what preserves the collision you want, and reversing it silently admits duplicate live accounts.

Follow-up

A deleted account re-registers with the same address the next day. Do the old resource rows follow the new user_id, and how does the API keep the two principals apart?
How do you honour an erasure request while resource_revision.actor_user_id still references this table?

What programming languages are you proficient in, and how have you applied them in your projects?

Medium · Technical / Domain Questions

Approach

Name two or three languages and rank them honestly (daily use, shipped production code, read-only). Lead with the ones that match the requirements candidates describe for this role: Python, Java or C++.
For each language, give one project and the reason that language fit it, such as Python for data processing and tooling, Java for a long-running service, or C++ where memory layout and latency mattered. Then say what you personally built.
Show depth with one concrete language feature you relied on, for example Python generators to stream a large file in constant memory, Java's ExecutorService for bounded concurrency, or C++ RAII and move semantics to manage resources without leaks.
Name a trade-off or a mistake you hit in that language, such as Python's mutable default argument, the GIL limiting CPU-bound threads, or a dangling reference in C++, and how you fixed it.
Close with what you would pick for a new task and why. That shows you choose by constraints rather than habit.

Follow-up

How does memory management differ between Python (reference counting plus a cycle collector), Java (tracing garbage collection) and C++ (deterministic destruction)?
In Python, when would you use multiprocessing or asyncio instead of threads?
Which language feature have you misused, and what did it cost?

Describe a technical challenge you faced and how you resolved it.

Medium · Technical / Domain Questions

Approach

Pick a problem you diagnosed and fixed yourself, and open with the symptom and the constraint, for example intermittent data corruption under load with no reproduction locally.
Walk through the diagnosis as hypotheses and evidence: what you measured or logged, which hypothesis that ruled out, and the experiment that confirmed the root cause.
State the options you weighed (quick patch, redesign, revert) and why you chose one, including what it cost in time or complexity.
Give the measured result (error rate, latency, crash count before and after) and what you left behind to stop a repeat: a regression test, an alert, a design note.
Say "I" for your own actions and credit others by role. A story where your part is unclear does not answer the question.

Follow-up

What was the alternative you rejected, and would it have worked?
How did you confirm the fix addressed the root cause rather than the symptom?
What would you do differently if the same problem appeared tomorrow?

How would you approach optimizing a slow-running application?

Medium · Problem-Solving / Case Studies

Approach

Define slow before touching code: which operation, which percentile (p50 vs p99), since when, and for which users. A regression since a release points to a diff; slowness that grows with data points to complexity or a missing index.
Measure, then profile. Use cProfile or py-spy for Python, async-profiler or JFR for Java, and perf for C++, plus request tracing to split time between CPU, database, network and lock waits.
Fix the largest share first, and use Amdahl's law to justify it: speeding up a fraction p of the runtime by a factor s gives an overall speedup of 1 / ((1 - p) + p / s).
Know the usual culprits: N+1 queries, missing indexes (check with EXPLAIN ANALYZE), quadratic loops, unbatched network calls, missing connection pooling, oversized payloads and GC pressure. Name the fix for each: batching, an index, a dict lookup, caching with an invalidation rule.
Re-run the same benchmark under the same load after each change, keep a latency budget, and add a performance test or alert so the regression cannot come back unnoticed.

Follow-up

The profiler shows time spread evenly across many functions. What do you do next?
When does adding a cache make things worse?
How do you keep a performance fix from regressing six months later?

If you encountered a significant bug just before a product launch, what steps would you take?

Medium · Problem-Solving / Case Studies

Approach

Reproduce the bug first and size it: who is affected, how often, and whether it touches safety, security, data integrity or money, or is cosmetic. That classification drives every later step.
Tell the release owner and stakeholders early, with facts: what breaks, for whom, the workaround if any, and your confidence in a fix estimate.
Lay out the options with their risk: a minimal targeted fix with a regression test, disabling the feature behind a flag, shipping with a documented workaround, or delaying the launch. A large last-minute change carries its own risk.
If you fix it, keep the diff small, add a test that fails without the fix, and run the critical-path regression suite before go/no-go. Plan the rollback before you deploy.
Afterwards, write a blameless review of why testing missed it, and add the missing test or check to the pipeline.

Follow-up

The product owner wants to ship anyway. How do you make the case either way?
How would you decide between a fix-forward and a rollback after launch?
What change to your test process would have caught this earlier?

Explain how you would design a system to handle a sudden increase in user traffic.

Medium · Problem-Solving / Case Studies

Approach

Clarify the spike: how large and how fast (2x over hours vs 10x in minutes), whether reads or writes dominate, and what degraded behaviour is acceptable. Then set targets for p99 latency and error rate.
Keep the app tier stateless behind a load balancer so it can scale horizontally, with sessions in a shared store, and autoscale on CPU or request rate. Autoscaling reacts with a delay while instances boot, so keep headroom or scale ahead of known events.
Take load off the database: a CDN for static content, a cache such as Redis for hot reads (jittered TTLs and request coalescing to avoid a stampede), read replicas, and connection pooling so new app instances do not exhaust DB connections.
Move non-urgent writes and side effects onto a queue so the spike is absorbed and processed at a steady rate, with idempotent consumers because messages can be redelivered.
Protect the core path: rate limiting (token bucket) per client, load shedding and timeouts with circuit breakers on dependencies, and graceful degradation such as switching off non-essential features. Validate it all with a load test before the real event.

Follow-up

What fails first at ten times normal traffic, and how would you know from your dashboards?
How do you stop a cache stampede when a hot key expires?
Writes cannot be queued because users need immediate confirmation. What changes?

Every query on one table stalls for forty seconds mid-deploy

Medium · lock contention · ddl · migrations

During a release on PostgreSQL, every query touching resource times out for about 40 seconds and then recovers with no intervention. The release ran one migration, ALTER TABLE resource ADD COLUMN archived_reason TEXT, and the migration log shows it completing in 6 ms. Unrelated tables showed no change in error rate. Explain how a 6 ms statement caused a 40-second stall, give the ordered checks you would run on a live system to confirm it, and give the migration procedure that prevents a repeat.

Approach

Separate the statement's duration from the lock's duration. ADD COLUMN with no default is a catalogue-only change and genuinely runs in milliseconds, but it requires ACCESS EXCLUSIVE, and it cannot acquire that until every transaction already touching the table has finished.
Account for the queueing, which is the part that surprises people. A lock request that is waiting blocks later requests for conflicting modes behind it rather than letting them overtake, so one long-open transaction holds the DDL and the DDL holds all the traffic. The stall length is set by the longest open transaction, not by the size of the change.
Confirm on a live system in this order: pg_stat_activity for that table ordered by xact_start, looking for the oldest transaction and specifically for state = idle in transaction; then pg_locks where granted = false to find the waiter; then join them on pid to name blocker and blocked. pg_blocking_pids() does that join for you and is the fastest single call.
Prevent rather than merely time it better. Set lock_timeout to a second or two on the migration session so the DDL abandons the queue after a bounded wait and is retried, instead of holding it for as long as the oldest transaction lives. Be exact about what that buys: queries arriving during the wait still queue behind the pending ACCESS EXCLUSIVE request, so each attempt costs them up to one lock_timeout of added latency. The outage goes from 40 seconds to about one second per attempt, not to zero. Also run migrations away from deploy-time peaks, and put a statement timeout and an idle-in-transaction timeout on the analytics role that opens the long transactions.

Follow-up

The same release also wants NOT NULL on that column. What is the sequence that gets there without a long lock?
Your lock_timeout retry fails ten times in a row because the analytics transaction is always open. What do you change?

Suggested practice checklist

The plan alternates spoken story practice with timed technical practice, following the order of the reported rounds. Each day names the reported questions it covers so you can check them off.

Day 1 — Map the loop and inventory your material

Write the three reported rounds (Phone Screening, Technical Assessments, In-Person Interviews) on one page and mark the one you are least ready for.
List your last three projects with the language, your role, one hard decision and one result for each.
Match each language you list (Python, Java, C++) to the project where you used it.

Deliverable: A one-page loop map plus a project inventory with language, role, decision and result for each project.

Day 2 — Phone screening: languages and lifecycle

Write and say aloud your answer about the languages you are proficient in, with one project per language.
Write and rehearse your lifecycle and methodology answer using one real sprint or release as the example.
Write a short resume walkthrough and three questions for the interviewer.

Deliverable: Two spoken answers (languages, lifecycle) and a resume walkthrough with three questions to ask.

Day 3 — Stories: technical challenge and difficult teammate

Outline the technical challenge story: symptom, diagnosis steps, fix, how you confirmed it and what you learned.
Outline the difficult teammate story with the conversation, the agreement and your own part in it.
Record both answers aloud and cut anything that blames or that lacks a result.

Deliverable: Two written story outlines and two recorded spoken versions.

Day 4 — Stories: leading and prioritising

Outline the lead story with the goal, the hard decision, the outcome and what you would change.
Write your prioritisation method as four steps, then attach one real episode of competing deadlines.
Prepare a persuasion story and a feedback story as extra material.

Deliverable: A lead story, a prioritisation method with episode, and two short backup stories.

Day 5 — Coding in Python and the session data structure

Solve two or three Python problems on hash maps, arrays or linked lists, stating time and space complexity aloud.
Implement a session store with create, lookup, refresh and expiry, and explain how expired sessions are removed.
Write a two-sentence complexity explanation for each solution without looking at notes.

Deliverable: Working Python solutions with complexity statements, and a session store with its design notes.

Day 6 — Diagnosis, launch bug and traffic spike

Rehearse a spoken diagnosis for a slow application using the order reproduce, measure, isolate, fix, verify.
Outline the steps for a serious bug found just before a launch: triage, workaround or rollback, fix and communication.
Outline a design for a sudden jump in traffic with caching, load balancing, queues and degraded modes.

Deliverable: Three spoken outlines (slow application, launch bug, traffic spike) and notes from the practice drill.

Day 7 — Full rehearsal

Run a mixed mock: one language answer, one story, one diagnosis, one coding problem, in a random order.
Re-record your weakest answer from days 2 to 6 and compare it with the first version.
Finalise your questions for each kind of colleague you may meet in person.

Deliverable: A recorded mock session, one re-recorded answer and a final list of questions to ask.

Stories to prepare for Brunswick

Candidates report that the phone screening covers your background and fit, and that the in-person stage consists of several interviews with team members about collaboration and communication. Both reward specific stories about your own work. Rehearse each prompt with a real situation, a decision you made and an outcome you can describe honestly.

Can you describe a time when you had to persuade a team member to adopt your idea?
How do you handle feedback or criticism from peers?

Can you explain the software development lifecycle and your experience with different methodologies?

Medium · Technical / Domain Questions

Approach

Open with one sentence on the lifecycle itself: gather requirements, design, implement, test, release, maintain. Then say which model your teams actually used. This is an experience question wearing a definitions question, so do not spend the answer reciting textbook phases.
Compare two approaches you have worked in, for example staged hand-offs against Agile sprints with planning, daily check-ins and retrospectives. For each, say what it did well (feedback speed, handling changed requirements, documentation) and what it cost.
Anchor in one real project: where requirements came from, how work was split into tickets, how code was reviewed and tested, how releases went out and how bugs returned. Code review and sprint planning are named in the responsibilities candidates describe for this role, so those details are relevant.
Show judgement about fit. Firmware that is hard to update after shipping pushes toward more up-front design and test; a web analytics dashboard tolerates frequent small releases. Only claim methods you have used.

Follow-up

What did a retrospective change on your team, and how did you check it helped?
How do you handle a requirement that changes in the middle of a sprint?
Who owned testing on your last project and where in the process did it happen?

Tell me about a time you had to work with a difficult team member. How did you handle it?

Medium · Behavioral / Leadership

Approach

Choose a real work conflict (missed hand-offs, contentious reviews, unclear ownership), not a personality complaint. Describe the behaviour and its effect on the project in two neutral sentences.
Show the first thing you did: a private, direct conversation in which you asked for their view and listened for the constraint behind the behaviour, such as workload or unclear scope. Say what you learned.
Describe the concrete agreement that came out of it: a review turnaround, a written interface, a shared checklist, pairing on the hard part. Concrete changes make the story checkable.
Mention escalation only as a later, deliberate step, framed as asking a manager for help with process rather than reporting a person.
Give the result honestly, including partial success and what you would change in your own behaviour. The mistake that sinks this answer is casting the other person as a villain.

Follow-up

What if the conversation had changed nothing?
At what point do you involve a manager, and how do you describe the problem to them?
What did you learn about your own part in the friction?

Describe a situation where you took the lead on a project. What was the outcome?

Medium · Behavioral / Leadership

Approach

Pick a project where you carried the responsibility, even without a title, and say how you came to lead it: assigned, volunteered or filling a gap. Give the goal, team size, constraints and deadline once, then move to actions.
Describe leading actions rather than personal coding: breaking the goal into work items, assigning by strengths, setting a check-in rhythm, removing blockers and keeping stakeholders informed.
Include one hard decision with its trade-off, such as cutting scope to hold a date, resolving a design disagreement or raising a risk early. This is the part that shows leadership.
State the outcome with whatever evidence you honestly have (what shipped, what changed for users or the team) and credit the team. Use "I" for your decisions and "the team" for shared work, and never claim results you did not drive.
Close with one thing you would do differently. A story with no outcome, or one where you silently did everything yourself, does not answer the question.

Follow-up

How did you handle a teammate who fell behind?
What did you do when two people disagreed about the design?
How did you keep product managers or stakeholders informed as plans changed?

How do you prioritize tasks when faced with multiple deadlines?

Medium · Behavioral / Leadership

Approach

State your method before the example: list every commitment with its date, estimate effort, rank by the cost of missing it and by what blocks other people, then start with the item whose delay hurts most. Revisit the ranking when something changes.
Separate urgent from important, and test whether anything can be cut, shortened, delegated or renegotiated before you decide something will slip.
Say you raise conflicts early. Tell the people waiting on the work that two dates collide and offer options (move a date, reduce scope, add help) rather than silently missing one.
Support the method with one real episode: the competing deadlines, how you ranked them, what you said to whom and how it ended. If an estimate was wrong, say how you adjusted.
The answers that fail are "I work longer hours" and "everything is top priority", because neither shows a decision.

Follow-up

What if two managers each insist their task is first?
How do you handle a new urgent request in the middle of a sprint?
How do you account for estimates that turn out to be wrong?