The five reported questions are broad fundamentals that you explain in your own words: the difference between p and div, how you test and deploy in production, a Django concept from your past projects such as overriding methods, trade-offs between database architectures under heavy traffic, and how you keep code quality up. All five sit in the technical and domain knowledge group, so the preparation is spoken answers backed by examples from systems you have actually shipped, not memorised definitions.
Candidates also describe a coding assessment and topics such as algorithmic thinking and problem solving, with memory management, performance, best practices in your main language and testable, modular code as areas to review. Reported example scenarios include refactoring a legacy function and listing the common pitfalls of distributed systems. For the second, be ready to name partial failure, retries that need idempotency, duplicate or out-of-order delivery, clock differences between machines and timeouts, and say what you do about each.
The architecture areas candidates mention are API design and communication patterns, data consistency and state management, and scalability for high-availability systems. Because the product schedules, plays and measures content on screens, it helps to have a basic picture of how a fleet of remote devices could fetch instructions and report back, and to reason about what happens when a device is offline. Treat that as a practice exercise for your own reasoning, not as knowledge of Broadsign's internals.
Candidates report team-wide meet and greets and say collaboration matters alongside technical skill. The stated responsibilities include working with product managers, doing code reviews, taking ownership and spotting technical debt, so prepare stories on each of those and a few questions of your own for the team.
Preparation focus
editorialCandidates report that the process starts with a recruiter screen covering background, salary and interest, then moves to a technical evaluation. That evaluation mixes online coding assessments with technical deep-dives, in person or remote, and team-wide meet and greets with developers, architects and team leads. The order and number of stages are not reported, so ask your recruiter which stages you will face, which language the coding assessment allows and whether any problem will be solved on paper. Prepare in three layers: spoken answers for the five reported fundamentals questions, timed coding and refactoring practice, and stories showing collaboration and ownership.
What to demonstrate
- Command of your main language and computer science fundamentals, including memory management and performance
- Ability to write testable, modular code and to refactor existing code safely
- Architectural thinking: API design, data consistency and scalability
- Communication and collaboration, reported as important in the meet and greets
How to prepare
- Rehearse spoken answers to the p versus div, testing and deployment, Django method overriding, database architecture and code quality questions, each with an example from your own work
- Practise a refactor of a legacy function from your own code: add tests that pin current behaviour first, then restructure in small steps
- Work PracHub's coding drills with a stated complexity for each solution, and solve one on paper while explaining your logic aloud
- Write your list of distributed systems pitfalls with a mitigation for each, and prepare two or three stories about code review, ownership and working with product managers
PracHub editorial advice for the preparation topics above.
Answering p versus div with only "block versus inline"
Both are block-level by default, so that answer is wrong. The real difference is meaning: p marks a paragraph of text and carries semantics for browsers, screen readers and search engines, while div is a generic container with none. A p cannot contain block-level elements, because the parser closes it early. Reach for div (or a more specific semantic element) when grouping for layout.
Describing a Django override without the pitfalls
Pick a real example, such as overriding save() on a model. Show that you call super().save(*args, **kwargs), and say what the override misses: bulk_create and queryset update() skip save(), and signals still fire separately. Explain why you chose an override over a signal, a manager method or a form clean step.
Answering the database question with "SQL is consistent, NoSQL scales"
Start from the access pattern and the consistency you need. Compare a relational store, a document store and a key-value store on joins, transactions, schema change and horizontal scaling, then add read replicas, caching and sharding, and name what each costs you, such as replica lag or cross-shard queries.
Stopping the testing and deployment answer at "I run unit tests"
Walk through the path from commit to production: unit and integration tests, review, automated build, staged or canary release behind a flag, monitoring after release, and a rollback plan. Mention that schema migrations must stay compatible with the previous version of the code while both run.
Solving the coding or paper problem in silence
Candidates are told to think aloud and to treat a paper problem as a conversation. State your approach and its complexity before writing, name the edge cases, then write. If you do not know something, say how you would find it out.
Choose a category, try a prompt, then open its approach, worked solution or follow-up when you need it.
Track a rolling failure rate per destination for circuit decisions
The egress service delivers about 1,500 webhooks per second across roughly 40,000 destinations, each call bounded by a 10 second timeout. Maintain, per destination, the failure rate over the trailing 60 seconds so a caller can ask before dispatch whether the circuit should open. Attempts arrive as (destination_id, finished_at_ms, outcome). Requirement: amortised O(1) per attempt, with total memory bounded by the destination count rather than by traffic. Give the structure, its exact memory, and the rule that stops a destination with three attempts from opening a circuit.
Approach
- Name the exact-deque version and then reject it as the default. Holding timestamps and advancing a tail pointer past anything older than now minus 60 seconds is a correct two-pointer window at amortised O(1) per attempt, but its memory tracks in-window traffic, so one destination in a retry storm holds hundreds of thousands of entries while thousands of quiet destinations hold none.
- Use a ring of 60 one-second buckets per destination, each bucket a pair of counters for attempts and failures. On an attempt, advance the ring by the elapsed whole seconds, zeroing at most min(elapsed, 60) buckets, then increment the head. That is amortised O(1) with a fixed footprint per destination.
- State the footprint: 60 buckets times two 4-byte counters is 480 bytes of payload per destination, so 40,000 destinations is roughly 20 to 25 MB with per-entry overhead, bounded by the catalogue rather than by the rate. The cost is granularity, since the oldest bucket ages out in whole seconds, which is far tighter than the decision needs.
- Require a minimum sample before the circuit may open. A destination with three attempts and three failures reads as 100 percent and is not evidence; a floor of roughly 20 attempts in the window makes the ratio meaningful, and below that floor use a run of consecutive failures as the trigger instead.
Follow-up
- The fleet is 30 instances and each sees roughly a thirtieth of a destination's traffic. Where does the rate actually live, and what does a per-instance answer get wrong?
- A destination answers in 9.5 seconds and succeeds. It is not failing but it is consuming your per-destination concurrency. What signal should open the circuit here?
Merge partitioned event streams into one ordered feed with bounded lateness
The read-model service consumes 64 log partitions carrying about 4,000 events per second in total. Each partition is ordered within itself, but partitions drift by up to 30 seconds, and the activity feed must present a tenant's events in occurred_at order. Produce the merge. State its complexity, the buffer it requires in events and in bytes, what happens when one partition is idle, and what you do with an event that arrives after you have already emitted its position. Payloads average 1 KB.
Approach
- Merge with a min-heap over the 64 partition heads keyed on (occurred_at, event_id): O(log P) per event and O(n log P) overall. The tie-break on event_id is what makes the output deterministic when two partitions carry the same millisecond, which matters because the feed is paginated and a non-deterministic order reorders pages under the reader.
- Emitting the heap head is only correct once every partition has produced everything up to that timestamp, so the emit condition is a watermark: the minimum across partitions of the highest occurred_at seen, less the allowed lateness. Events are held until the watermark passes them, which is what turns individually ordered streams into a jointly ordered one.
- Size the buffer from the lateness rather than guessing: 4,000 events per second times 30 seconds is 120,000 buffered events, and at 1 KB each about 120 MB of heap. That number is the real price of the ordering guarantee and belongs in front of whoever asked for it.
- Handle the idle partition explicitly, because it fails the feed rather than corrupting it: a partition with no traffic never advances its own maximum, so the watermark freezes and output stops entirely. Either every partition emits a periodic idle marker carrying the broker's current time, or the watermark falls back to wall clock for a partition silent beyond a threshold.
Follow-up
- The lateness budget is raised to five minutes. What is the new buffer, and what besides memory changes?
- The consumer restarts. Where does it resume from, and what does the feed look like for the first 30 seconds?
Diff a projection against the primary without per-row point reads
The listing projection has drifted and some rows show a stale version. The primary holds 40,000,000 resource rows across 12,000 tenants while serving 1,200 writes and 14,000 reads per second. The obvious repair, reading each resource row and comparing its version against the projection, is correct and would eventually finish. Explain precisely why it is unacceptable here, then give a diff that finds the differing rows, state its complexity, and make it safe to run against a live primary. Replication lag is usually under 100 ms and is not bounded.
Approach
- Quantify the naive cost rather than calling it slow: 40,000,000 point reads at even 0.5 ms each is over five hours serialised, and the only lever is concurrency, which is exactly what you cannot spend. The primary's pool is sized for the write path, and 40,000,000 random reads evict the buffer cache that sustains the 85 percent cache hit rate, so the audit degrades the system it is auditing.
- Replace random access with one ordered pass per side. Both sides can be read in (tenant_id, resource_id) order, which is a sequential scan on each and a merge join in O(n) time and O(1) memory. For a dense diff that is the whole answer, and it reads the primary once instead of 40,000,000 times.
- For the expected sparse case, compare range hashes instead of rows: partition the key space, compute per range an order-independent aggregate over hash(resource_id, version), compare aggregates, and descend only into ranges that differ. With d differing rows and branching factor B, at most d ranges mismatch per level, so the drill-down examines O(d log_B(n/d)) ranges and reads full rows only in mismatching leaves.
- Aggregate with a sum modulo 2^64 or a multiset hash, never XOR. XOR is order-independent but self-cancelling, so two rows wrong in the same way, or a row duplicated on one side, leave the range aggregate matching and the range is declared clean.
Follow-up
- The diff reports 900 stale rows. How do you decide between patching those rows and rebuilding the projection from resource_revision?
- Same job, but the projection lives in a search index that cannot be scanned in key order. What changes?
Stop tag and share joins from fanning out a page
resource_tag is (resource_id, tag_id) with PK (resource_id, tag_id); resource_share is (resource_id, shared_with_user_id, permission). The tagged-and-shared listing inner-joins resource to both, filters tenant_id, tag_id = ANY($2) and shared_with_user_id = $3, orders by updated_at DESC and takes 50. Pages come back with fewer than 50 distinct resources and the total in the header is far too high. Explain the row multiplication, rewrite both the page query and the count query so each is correct, and name the index each one needs. PostgreSQL 16.
Approach
- Do the arithmetic against the predicates that are actually there. An inner join emits one row per matching child row, and both joins are filtered: tag_id = ANY($2) admits only the requested tags, shared_with_user_id = $3 admits one user's share rows. So a resource holding three of the requested tags and shared with $3 once yields three rows, not one — the multiplier is its count of matching tags times its share rows for that single user, and that second factor is 1 unless the table admits duplicate (resource_id, shared_with_user_id) pairs. LIMIT 50 then limits rows rather than resources, and COUNT(*) counts pairs — the header is the product, not the population.
- Reject DISTINCT as the fix. It deduplicates after the product has been built, so the planner must materialise and sort the fanned-out set before the LIMIT can apply, and it leaves any SUM or AVG in the same select list wrong.
- Rewrite both filters as semi-joins, keeping resource as the only row source: AND EXISTS (SELECT 1 FROM resource_tag rt WHERE rt.resource_id = r.resource_id AND rt.tag_id = ANY($2)) and the same shape against resource_share. A semi-join stops at the first match per resource and preserves the driving index order, so ORDER BY updated_at DESC, resource_id DESC LIMIT 50 still stops after 50 rows.
- Count with the same predicates and no join at all: SELECT count(*) FROM resource r WHERE r.tenant_id = $1 AND r.status = 'active' AND EXISTS (...) AND EXISTS (...). Nothing multiplies a resource, so the number is the population.
Follow-up
- The filter changes from 'any of these tags' to 'all of these tags'. Rewrite it and state what it costs relative to the ANY form.
- A resource can be shared with the same user twice under different permissions. Does your count change, and should it?
Write the update path that detects a concurrent edit
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?
Explain the difference between HTML elements like `<p>` and `<div>`.
Explain the difference between HTML elements like <p> and <div>.
Approach
- Lead with semantics: is a paragraph of text, and its content model allows phrasing content only (text, , , , ). is a generic container with no meaning, and it can hold any flow content, including other s, lists and tables.
- Separate meaning from rendering. Browser default stylesheets make both display: block; also gets a default top and bottom margin of 1em and gets none. CSS can change how either one displays, but it cannot change what the element means or what it may contain.
- Explain the parser edge case. A end tag is optional, so a block-level start tag such as , or closes an open implicitly. Markup like x ends up as x: the stray creates a second, empty paragraph.
- Cover accessibility. Assistive technology exposes as a paragraph, while gets the generic role. Use only for layout or styling hooks, and pick a semantic element (p, section, article, nav, button) when the content has a real meaning.
Follow-up
- What does the DOM look like if you put a or a inside a , and why?
- When would you choose or over a ?
- If you set a to display: inline, what changes and what stays the same?
How do you approach testing and deployment procedures in a production environment?
How do you approach testing and deployment procedures in a production environment?
Approach
- Lay out the test layers and what each one catches. Fast unit tests run on every commit. Integration tests run against a real database and real dependencies, for example in Docker containers. A small end-to-end smoke suite covers the main user paths. CI runs all of it on every pull request, and a failing run blocks the merge.
- Build one immutable artifact per commit, such as a container image tagged with the git SHA, and promote that same artifact from staging to production. Only configuration changes between environments, so production never runs a rebuilt binary nobody tested.
- Name a rollout strategy and its trade-off. A rolling update with readiness probes (a Kubernetes Deployment, for example) is simple. A canary sends a small share of traffic to the new version first. Blue/green costs double capacity but rolls back with one traffic switch. Watch error rate and latency against a baseline, and roll back automatically when they get worse.
- Handle schema changes with expand/contract. First ship a backward-compatible migration (add a nullable column, write to both columns), then ship the code that uses it, and drop the old column in a later release. Old and new versions run side by side during a rollout, so this order lets both work against the schema.
- Separate deploying from releasing. Feature flags let a risky feature go out dark and be switched off without a redeploy. After every deploy, run smoke checks, and if something reaches production anyway, add a regression test for it.
Follow-up
- How do you roll back a release that included a database migration?
- Which metrics and thresholds would make you halt a canary, and how long would you let it run?
- How would you test a bug that only shows up under production traffic or data volume?
Can you explain a complex concept from your past projects, such as overriding Django methods?
Can you explain a complex concept from your past projects, such as overriding Django methods?
Approach
- Pick one override you actually wrote and structure the answer: what the default behaviour did, why it was not enough, which hook you overrode, and what went wrong or nearly went wrong. A common example is Model.save() that fills in a derived field. Call super().save(*args, **kwargs) and pass the arguments through, so that options like update_fields and using still work.
- Show you know what the override does not cover. QuerySet.update() and bulk_create() never call save() and do not send pre_save/post_save signals. QuerySet.delete() does not call each model's delete(), although it still sends pre_delete/post_delete for every deleted object. Model.save() does not run full_clean(); ModelForm validation does.
- For class-based views, explain the usual hooks. Override get_queryset() to scope objects to request.user, get_context_data() (call super() and extend the dict it returns) to add template context, and form_valid() to attach the current user before saving. If you used Django REST Framework, perform_create() plays the same role.
- Compare the alternatives and say why you chose this one. Overriding save() keeps the logic next to the model but is easy to bypass. Signals decouple apps but hide control flow. A custom Manager or QuerySet method gives callers one explicit entry point. Validation belongs in clean() or the form, not in save().
- Close with how you tested it: a unit test for the overridden path, and a test or code-review rule that catches bulk operations skipping the logic.
Follow-up
- What happens to your save() logic when someone runs a queryset.update() on that model?
- Why did you choose an override over a post_save signal or a custom manager method?
- How would you make the derived field stay correct after a bulk import?
What are the trade-offs between different database architectures in a high-traffic system?
What are the trade-offs between different database architectures in a high-traffic system?
Approach
- Clarify the workload before naming any technology: read/write ratio, peak QPS, data size and growth, the queries that must be fast, and which data needs strong consistency (money, schedules) versus data that can be eventually consistent (counters, event logs).
- Single primary with read replicas: reads scale out, writes still go to one node, and asynchronous replicas lag. That breaks read-your-writes, so send a user's reads to the primary for a short window after they write, or read from a replica only once it has caught up to the write's log position.
- Sharding (horizontal partitioning): writes and storage scale out, but the shard key decides everything. Queries that do not include the key fan out to every shard, cross-shard joins and transactions get expensive, a hot key overloads one shard, and resharding later means migrating data. Consistent hashing limits how much data moves when nodes change.
- Relational versus key-value or wide-column stores (DynamoDB, Cassandra): the NoSQL stores scale writes horizontally and offer tunable consistency, but you model tables around known access patterns, denormalize, and give up ad hoc joins. In CAP terms, a partition forces a choice between consistency and availability; PACELC adds the latency-versus-consistency choice when there is no partition.
- Mention the supporting layers: a cache-aside Redis tier, with an invalidation story and stampede protection. A separate analytical store, fed by change data capture or an event stream, so heavy reporting never runs on the OLTP primary. Multi-primary setups need a conflict-resolution rule, and last-writer-wins silently loses concurrent updates.
Follow-up
- How do you guarantee read-your-writes when reads go to asynchronous replicas?
- Pick a shard key for a high-volume event table. Which query does it make expensive, and what happens with a hot key?
- At what point would you move a workload off a single Postgres primary, and what would you try before sharding?
How do you ensure code quality during the development lifecycle?
How do you ensure code quality during the development lifecycle?
Approach
- Walk through the lifecycle in order instead of listing tools. Before coding, agree on the interface and acceptance criteria for a non-trivial change. While coding, write small, testable modules: inject dependencies, keep pure logic away from I/O, and write tests alongside the code. Before merge, review. After deploy, monitor.
- Name the automated gates and what each one catches: a formatter and linter (for example ruff/black or ESLint/Prettier) for style and simple bugs, a type checker (mypy, TypeScript) for interface mismatches, and unit plus integration tests. All of them run in CI, and a merge needs every check to pass.
- Describe review as a practice. Keep pull requests small enough to read in one sitting, self-review the diff first, and review for correctness, error handling, test coverage of the changed behaviour, naming and security. On risky changes, the reviewer pulls the branch and runs it.
- Treat coverage as a signal, not a target. Look at coverage of the changed lines and at whether the tests assert behaviour, because a high percentage with weak assertions catches nothing. For legacy code, write characterization tests before refactoring so the refactor can be checked against the current behaviour.
- Close the loop after release. Error tracking and dashboards catch regressions, every escaped bug gets a regression test, and recurring problems become a lint rule or a review checklist item rather than a reminder.
Follow-up
- How do you settle a code-review disagreement about design rather than style?
- Coverage is high but bugs still reach production. What do you change?
- How would you introduce linting and type checking to a large legacy codebase without one huge reformatting PR?
Edge instances grow 400 MB per hour until the nightly restart
Edge API instances start at 700 MB resident and grow about 400 MB/hour; a nightly rolling restart has hidden it for weeks. Growth continues unchanged when request rate halves overnight, p99 degrades in the last hours before an instance is recycled, and heap used immediately after a forced full GC rises monotonically. The service holds no product state. Name the discriminating measurement that separates the plausible causes, give the most likely cause, and give the fix and how you would verify it.
Approach
- Separate resident memory from live heap first, because they fail differently. Resident size can grow from fragmentation, native buffers or thread stacks while the heap is flat; heap used after a full GC rising monotonically is the measurement that says objects are reachable and not being released. You already have it, so this is retention, not fragmentation, and that closes off half the candidate list.
- Use the rate's independence from traffic as the discriminator. Growth that continues at half the request rate rules out per-request objects that are merely slow to collect and points at a structure that grows with distinct values observed rather than with call volume. Write the candidates that have that property: a metrics registry keyed on a high-cardinality label, an unevicted cache, an interner, a per-key lock map.
- Take two heap snapshots an hour apart and diff by retained size, reading the dominator tree, not by allocation count or instance count. Expect one root holding a map with millions of entries, then follow the reference chain to the code that inserts and never removes. Allocation profilers point at churn, which is the wrong signal here.
- The candidate that fits this service is an observability label carrying an identifier, such as a request path recorded before templating so that /v1/resources/48213 becomes its own metric series. That grows with distinct ids seen, is independent of rate, and explains the late p99 degradation, since GC cost rises with the size of the live set.
Follow-up
- Post-GC heap is now flat but resident size still creeps. What are you looking at, and does it matter?
- How would you have detected this before an OOM, given the nightly restart masked the trend?
The plan starts by confirming the format, then works the five reported questions, the coding assessment and architecture topics, and finishes with collaboration stories and a full dry run. Days point at PracHub practice drills where they fit.
Prepare, practise & reflect
One practical outcome each day. Spend longer where you need it.
0 / 7 done01Confirm the format and map the reported topics
- Write to your recruiter asking which stages you will face, the language and platform for the coding assessment, and whether a problem may be solved on paper
- Write the five reported questions on one page and put a one-line outline answer under each
Deliverable: A sent recruiter message and a one-page map with an outline answer for each of the five reported questions.
02HTML elements and Django method overriding
- Answer the p versus div question aloud: default display, content model, semantics and accessibility
- Choose one Django override from your own work, write its code and list what bypasses it (bulk_create, queryset update) and what you would use instead
Deliverable: A recorded spoken answer on p versus div and a written Django example with its pitfalls.
Practice prompt ↗Practice prompt ↗03Database architectures and SQL
- Build a comparison table of relational, document and key-value stores against transactions, joins, schema change, scaling and consistency, plus replicas, caching and sharding
- Do the PracHub SQL drills on join fan-out and detecting a concurrent edit, then tie each back to the trade-offs in your table
Deliverable: A filled comparison table, a spoken answer to the database trade-offs question and two solved SQL drills.
Practice prompt ↗Practice prompt ↗Practice prompt ↗04Testing, deployment, code quality and memory
- Write the pipeline from commit to production for a project you shipped, naming test layers, release method, monitoring and rollback
- Take a legacy function of your own, add characterisation tests, then refactor it in small steps
- Work the PracHub debugging drill on edge instances growing in memory and list your investigation sequence
Deliverable: A written pipeline, a before-and-after refactor with its tests and a memory investigation checklist.
Practice prompt ↗Practice prompt ↗Practice prompt ↗05Coding assessment practice
- Solve the three PracHub coding drills, stating approach and complexity before you write each
- Solve one of them on paper, explaining your logic aloud as you go
Deliverable: Three solutions with stated time and space complexity, one of them handwritten.
Practice prompt ↗Practice prompt ↗Practice prompt ↗06Distributed systems and API design
- Write the pitfalls of distributed systems (partial failure, retries, duplicates, ordering, clock differences, timeouts) with one mitigation each
- Sketch an API for remote devices that fetch a schedule and report back playback, covering offline devices, retries and data consistency
Deliverable: A pitfalls-and-mitigations list and a one-page API sketch with its failure cases.
07Collaboration stories and full dry run
- Write three stories: a code review that changed your approach, a bottleneck or debt you raised yourself, and a disagreement with a product manager or teammate
- Rehearse the PracHub behavioral drills on arguing against a design and narrating an outage you owned
- Run one full mock covering a spoken fundamentals answer, a coding problem and a story, then write the questions you will ask the team
Deliverable: Three written stories, a completed mock and a list of questions for the team.
Practice prompt ↗Practice prompt ↗Expand any day for tasks and deliverables. Your progress is saved on this device.
Candidates report team-wide meet and greets and say collaboration counts alongside technical skill. The responsibilities they describe include syncing with product managers, reviewing code, taking ownership of tasks and flagging bottlenecks or technical debt, so rehearse a story for each. Use a situation, your action, the result and what you would change.
Argue against a design, lose, and commit anyway
Describe a design you argued against and lost. State the failure you predicted as a named mechanism, not a feeling about complexity: two services that would need one transaction, a projection with no rebuild path, a write path with no idempotency key. Say what evidence you brought, what the decision maker weighed instead, and what you did after the decision was made: what you instrumented, what you wrote down, and whether the prediction came true. Five minutes.
Approach
- State the prediction in falsifiable form up front: the mechanism, the condition that triggers it, and the observable outcome. A prediction that cannot be checked also cannot be credited to you later.
- Show the evidence you had at the time and label each piece honestly as measured, analogous, or intuition. Keeping the intuition is fine; disguising it as data is the thing that erodes your standing in the next argument.
- Represent the opposing case at full strength, including the constraint you did not control: a fixed date, a team boundary, or the fact that the decision was cheap to reverse and yours was not.
- Make disagree-and-commit concrete. Name the artefact you left behind so the prediction could be settled without you: the alert and its threshold, the counter on the dashboard, the decision note that recorded the trade-off and the condition that would revisit it.
Follow-up
- What threshold on that alert would have proved you right, and did anyone ever look at it?
- If the same proposal arrived tomorrow with the same deadline, would you argue it the same way?
Reverse your own decision and price the reversal
Describe a technical decision you made and later reversed. Pick one that cost something: a service you split and merged back, a cache you added and removed, an index you created that pushed the planner onto a worse plan, a projection you rebuilt from scratch. State what you believed when you decided, the measurement that changed your mind, how long the wrong version ran in production, and what the reversal cost in migrations, dual writes, and a deprecation window for callers you did not own.
Approach
- State the original rationale without irony, in the version you would still defend given what was known then. If it is not defensible, the story is about carelessness rather than judgement, and a different example serves you better.
- Give the measurement that moved with a before and after: the p99 that did not improve, the cache hit rate that sat at 40%, the plan that flipped to a sequential scan once the table passed a size you can name.
- Cost the reversal in steps, not adjectives: expand-and-contract deploys, the dual-write window, the callers who had to be notified, the rows already written in the wrong shape that had to be backfilled or abandoned.
- Distinguish reversal from rewrite by naming what you kept. Most good reversals preserve the schema or the interface and undo one decision inside it, which is also why they were affordable.
Follow-up
- What in that decision was irreversible, and did you know it was irreversible when you made it?
- How did you tell the people who had already built on top of the original decision?
Narrate an outage you owned from page to postmortem
Pick an incident you personally drove, ideally one where writes were affected rather than reads. In six to eight minutes: state the symptom as it first appeared on a dashboard, the blast radius you established before you knew the cause, the mitigation you applied and when, the mechanism you eventually proved, and the follow-up that would prevent a repeat. Bring numbers: error rate, tenants affected, minutes to mitigate, minutes to resolve. If you cannot name what you measured, choose a different incident.
Approach
- Open on the signal rather than the cause: which metric at which percentile moved, on which service, at what time, so the listener follows the same evidence you had rather than a conclusion you already reached.
- Separate mitigation from diagnosis out loud. State what you did to stop the bleeding (flag off, shed traffic, drain a lease, roll back a deploy) and say plainly that you did it before the mechanism was known, because those are two jobs with different deadlines.
- Establish blast radius in countable terms: how many tenants, how many writes, and crucially whether the effect was loss or only delay. An append-only revision table or a pending outbox row means the change survived and the projection was merely behind, which is a repair rather than a data-loss incident.
- Prove the mechanism instead of asserting it. Name the trace span that grew, the plan that flipped to a sequential scan, the lease that expired, plus one alternative you ruled out and the signal that stayed flat while you ruled it out.
Follow-up
- What would you do differently in the first five minutes, given the same dashboard and no more information?
- Which follow-up action did you deliberately not take, and why was dropping it the right call?
- 01
Tell me about a code review, given or received, that changed how you wrote code.
- 02
Describe a bottleneck or piece of technical debt you spotted yourself and how you got it addressed.
- 03
Tell me about a time you worked with a product manager to turn a vague requirement into something you could build.
- 04
Describe a production problem tied to a release and what you changed in testing or deployment afterwards.
- 05
Tell me about a time you did not know the answer and how you found it.
What does the Broadsign interview process look like?
Candidates report a recruiter screen about background, salary and interest, then a technical evaluation with online coding assessments and technical deep-dives, in person or remote, plus team-wide meet and greets. The order and number of stages are not reported, so ask your recruiter.
PracHub Software Engineer practice ↗Which questions do candidates report?
Five: the difference between HTML elements like p and div, your approach to testing and deployment in production, explaining a complex concept from past projects such as overriding Django methods, trade-offs between database architectures in a high-traffic system, and how you ensure code quality. Practise each aloud with an example from your own work.
PracHub Software Engineer practice ↗Will there be coding?
Candidates mention a coding assessment, and problem solving and algorithmic thinking appear among the reported topics. Ask which language and platform to expect and whether any problem is solved on paper. Prepare to state complexity and edge cases, and to explain your logic as you write.
PracHub Software Engineer practice ↗Do I need to know Django?
Django is a reported topic and one reported question asks about overriding Django methods. If you have not used it, be ready to explain a comparable framework hook from your own stack, such as overriding a model save method, and what bypasses it.
PracHub Software Engineer practice ↗How much should I know about the product?
Candidates are advised to learn the basics of digital out-of-home advertising: content shown on screens that has to be scheduled, played and measured. Cloud (AWS or GCP), Docker or Kubernetes and ad-tech experience are reported as nice-to-have, so mention them if you have them.
PracHub Software Engineer practice ↗How should I handle the meet and greets?
Candidates report that culture fit and collaboration matter and that these sessions work both ways. Bring stories about code review and teamwork, and prepare questions about team challenges, engineering culture and the roadmap.
PracHub Software Engineer practice ↗Sources & methodology 3 sources ↗
No official company page is cited. Rounds and questions come from candidate reports and PracHub editorial material; each source shows the date it was read.
- 01Broadsign Software Engineer candidate reports ↗
Company-reported rounds, questions and FAQ.
Candidate reports · Accessed 2026-09-22 - 02PracHub Software Engineer practice ↗
PracHub practice material, not company-reported.
PracHub page · Accessed 2026-09-22 - 03PracHub preparation framework ↗
PracHub preparation guidance.
PracHub page · Accessed 2026-09-22
Explore related engineering environments and compare the preparation focus.
Cognition Software Engineer Interview Guide
What Cognition asks Software Engineer candidates, from reported rounds and questions.
Vercel Software Engineer Interview Guide
Debugging practice for Vercel software engineer prep built on a reproduction you trust, narrowing to the smallest failing input, and changing one variable per run so the result stays attributable.
Together Ai Software Engineer Interview Guide
Failure-handling practice for Together Ai software engineer candidates: requests whose outcome a timeout leaves unknown, where a retry is safe, and deduplicating at the receiver instead of assuming exactly-once delivery.
Spotify Software Engineer Interview Guide
Timed practice for Spotify software engineer candidates who lose time to the clock: short problems taken end to end, then reviewed against a written rubric rather than a passing test run.