Interview guides · Barrick Mining

Barrick Mining Software Engineer Interview Guide

Barrick Mining Software Engineer interview guide: the five reported rounds, safety and leadership questions, answer outlines and a 7-day prep plan.

The 60-second brief

Barrick Mining is a mining company, and candidates describe the Software Engineer role as software for its plant and field operations. The work they list includes monitoring plant performance, managing resources and building tools that support on-site safety, with data pipelines for plant performance named as one example. Engineers are described as working alongside field engineers and operational managers across the full software lifecycle, from gathering requirements through testing, so the code ends up used by people on site.

This guide is for Software Engineer candidates preparing for the Barrick Mining loop. It covers the five reported stages, outlines for nine reported questions (risk and brittle failures, stopping unsafe work, your safety record, handling a missed deadline, leadership style, motivation for applying, your work history, plus two technical questions on performance metrics and applying your past experience), and a seven-day schedule that mixes behavioral rehearsal with reliability and design practice.

Candidates report five stages over roughly 4-6 weeks: Application Review, Virtual Screening, Recorded Video Assessment, In-Person Technical Panel and Final Offer Stage. Reports say the exact mix varies by role and region.

Focus areas

Behavioral Interviewing (Situational & Experience-based)
Safety Engineering / Unsafe Work Prevention
Leadership Principles & Being a Leader

Overview

Most of the reported questions for this loop are behavioral, and the themes are narrow: safety, risk, leadership and why you want this job. Candidates report that safety is the topic the conversation keeps returning to, so you need at least two stories where you identified a hazard and acted on it, plus an honest account of your own record. For software work, the closest honest material is usually a release you halted, a review gate you refused to bypass, an incident you escalated, or an on-call decision that protected users or operators.

The technical side is reported as practical rather than puzzle-driven: two design-flavoured questions ask which engineering metrics you would use to judge performance under high load, and how your former technical experience would carry over to the company's current infrastructure. Reports also mention remote or low-connectivity environments, failure handling and recovery, and the reasoning behind your architectural choices. Prepare to discuss a system you built in terms of what happens when a dependency disappears, a link drops or input arrives malformed, not only the case where everything works.

The role as described sits between software and site operations: field engineers and operational managers are the people you would build for, and you would translate between them and the engineering team. That makes communication part of the preparation. Practise explaining a technical trade-off to someone who does not write code, and tie each project on your resume to the operational outcome it served.

Candidates also report that the company's DNA-Core Values come up, with traits such as humble, transparent and taking ownership of your work. You do not need to recite them. Look them up, pick two, and make sure a story you already plan to tell demonstrates each one. A mining or industrial background is listed as a plus, not a requirement, so if you lack it, show that you understand why reliability and safety matter in that setting.

The Barrick Mining interview process

Candidates report five stages. The description of the recorded video assessment says you may be asked to complete one, and reports say the mix of virtual, recorded and in-person stages depends on the role and region.

Application Review

Candidates describe this stage as a review of your application for qualifications and fit. There is no live conversation yet, so the resume and any supporting text have to carry the argument. The requirements candidates report are proficiency in modern programming languages, system design experience and a record of collaborative development, with industrial or mining exposure, data analytics tools and remote or distributed systems listed as extras.

What to demonstrate

Whether your listed experience matches the reported requirements: modern languages, system design, collaborative development
Visible signals of reliability work, operational exposure or safety-conscious practice
Clarity of what you personally owned on each project

How to prepare

Rewrite the top of your resume so each bullet states what you built, what failed or was at risk, and the result in your own numbers.
Mark any bullet that touches the reported extras (industrial systems, analytics tooling, remote or distributed systems) and keep it prominent if it is true.
For every project listed, write down one obstacle and one trade-off you made, because reports advise being ready to go deep on each.

Virtual Screening

A virtual screening to look further at your background and skills. Candidates do not report a fixed format, so expect a conversation about your work history, your motivation and how your technical experience relates to the role. The background and motivation questions in this guide (your work history, why you applied, how your past experience applies to the company's infrastructure) are the natural practice material for a stage like this.

What to demonstrate

Background and skills as you describe them
Whether your experience connects to plant and field operations software
How clearly you explain your work and your reasons for applying

How to prepare

Build a short spoken version of your work history that ends on why this role fits, and rehearse it out loud.
Prepare the answer to why you applied using specifics from the role as reported: plant performance monitoring, resource management and safety tooling.
Pick one past system and rehearse how its design choices would carry over to a remote or low-connectivity operation.

Recorded Video Assessment

Candidates report that you may be asked to record a video assessment to show your abilities. A recorded format means you answer on camera without a live exchange, so nobody can steer you back on track and structure matters more than usual. Behavioral questions on safety, deadlines and leadership fit this format well, and the STAR method is the structure candidates are advised to use.

What to demonstrate

How well you structure an answer without interaction
Communication on camera
Concrete, personal examples rather than general statements

How to prepare

Record yourself answering the safety record and missed-deadline questions, then watch the playback for rambling, missing results and filler.
Write each story as Situation, Task, Action, Result in a few lines each and rehearse from those lines, not a script.
Check camera, audio and lighting before you start, and keep the stories you have not yet recorded in view as prompts.

In-Person Technical Panel

Candidates describe an in-person panel with technical peers and management that assesses technical skills. Reports on evaluation areas across the loop mention system design for reliability and scale, operational metrics, troubleshooting failing or brittle systems, and how software behaves in remote or low-connectivity conditions. A panel mixing peers and managers also suggests you will need to explain technical choices in business terms.

What to demonstrate

Technical skills in a panel setting, as reported
Reasoning behind architectural decisions
Handling of edge cases, failures and recovery rather than only the happy path
Communication across technical and non-technical listeners

How to prepare

Pick two systems you built and prepare a failure-mode walk-through for each: what breaks, how you detect it, how it recovers.
Prepare a metrics answer for high load: latency percentiles, throughput, error rate, saturation and queue depth, with the action each one would trigger.
Rehearse the leadership and deadline stories out loud with a colleague, since management may be in the room.
Practise explaining one trade-off in two versions, one for an engineer and one for an operations manager.

Final Offer Stage

Candidates describe this stage as a discussion of the final offer and next steps. Reports also mention background and safety checks as part of the overall process. Treat it as a conversation about terms and logistics rather than another evaluation, and come with questions about the role and team, which reports encourage.

What to demonstrate

Clarity about what you want from the offer and the role
Questions that show you understand the operational context
Readiness for the checks and next steps that follow

How to prepare

Write your questions about the team, the technology stack and how software and site operations work together.
Know your constraints on location, shifts or remote-site work in advance, since the soft skills reported include working in shift-based or remote environments.
Make sure any safety record or employment history you stated earlier is accurate and documented, because background and safety checks are reported.

Where candidates lose points

Telling a safety story where the decision was someone else's, or where you only reported a concern upward

Choose a story where you personally stopped the work. State the hazard, who could have been harmed, what authority you used, how you communicated it, and the process change that followed.

Describing every system you built only on its happy path, with no word on failure, recovery or poor connectivity

For each system you plan to discuss, prepare a failure-mode list: dependency outage, malformed input, partial write, overload, dropped link. Say how you detect each and how the system recovers or degrades.

Answering the engineering metrics question with a list of buzzwords such as CPU and latency, with no thresholds or actions

Name latency percentiles (p95 and p99, not averages), throughput, error rate, saturation and queue depth, then say what each one tells you and what you would do when it crosses a limit.

Claiming a spotless safety record in a sentence, or giving vague answers about it

State only what you can document, including one near miss or mistake and what changed, since background and safety checks are reported. Describe the record you actually have for your type of work.

Giving a generic reason for applying and a long-term goal that does not connect to plant and field operations software

Research the company's operations and its DNA-Core Values, then link a specific part of the role to something you have done and to where you want to grow in it.

Questions & practice

Archive a resource graph without breaking live references or recursing

Medium · graph traversal · topological order · tenant isolation

Resources reference other resources within a tenant; for the largest tenant the reference table holds up to 2,000,000 nodes and 8,000,000 edges. Archiving a resource must archive everything reachable from it that nothing outside the set still references, refuse when a live external referrer exists, and terminate when references form cycles, which they legitimately do. Produce the archive order and the refusal list, targeting O(V+E). Say what stops the traversal crossing a tenant boundary, and why recursion is the wrong control structure at this size.

Approach

Load the subgraph with the tenant predicate on both endpoints of the edge, not only on the side you started from. Scoping the left table alone is the classic cross-tenant leak: one mis-entered edge then pulls another tenant's resources into the traversal and, worse, into the archive.
Traverse iteratively with an explicit stack. A 2,000,000-node graph can hold a chain deep enough to exhaust a native stack in the low tens of thousands of frames, and that failure is a process crash rather than an error you can return.
Treat cycles as data rather than corruption: compute strongly connected components with Tarjan in O(V+E) using its own explicit stack, then condense. The condensation is a DAG, so a topological order over it gives the archive order, and every member of a component archives in one transaction because no order within a cycle is valid.
Decide refusals with reverse edges. A candidate is archivable only if every in-edge originates inside the candidate set, so build the transpose or count in-degrees restricted to the visited set, and emit each blocked resource with the id of the external referrer, which is the only part of the answer an operator can act on.

Follow-up

The graph is read in one query and the archive writes a minute later. What can change in between, and how do you make the write safe?
The candidate set is 400,000 resources. Is that one transaction, and if not, what does a half-finished archive look like to a reader?

Canonicalise a request body into a stable idempotency fingerprint

Medium · parsing · canonicalisation · hashing

idempotency_key.request_fingerprint is a SHA-256 over the method, path and canonicalised body, and a retry whose fingerprint differs must be rejected with 422 rather than served the stored response. Write the canonicaliser. Bodies are JSON up to 256 KB nested at most 32 levels; clients vary key order, whitespace and unicode escaping, and some send 64-bit ids as JSON numbers. Produce a deterministic byte string such that semantically identical bodies match and any semantic difference does not. State your complexity and name two normalisations you refuse to perform.

Approach

Parse once into a tree, then re-serialise under fixed rules: object keys sorted, array order preserved, one escaping convention, no insignificant whitespace. Parsing is O(n) and sorting keys is O(k log k) per object, so O(n log n) overall with O(depth) stack, and the 32-level cap is enforced during parsing because hostile nesting is how a canonicaliser becomes a stack overflow.
Sort keys by their UTF-8 bytes and say why the obvious implementation is wrong in some runtimes: a default string comparison that orders by UTF-16 code units places surrogate pairs, meaning code points from U+10000 up, below U+E000 to U+FFFF, which is not UTF-8 byte order, so two services written in different languages disagree on the same document.
Do not re-encode numbers through a double. IEEE-754 binary64 represents integers exactly only up to 2^53, so normalising a 19-digit id through a float changes it, and 1 against 1.0 cannot be reconciled without deciding whether they are the same value. Preserve the literal token, and require ids as strings at the API boundary if you want them comparable.
Reject duplicate keys rather than picking one. JSON permits them and parsers disagree, most keeping the last, so any choice you make ties the fingerprint to a parser detail that the code handling the request does not necessarily share.

Follow-up

A client sends the same logical request with an extra field your API ignores. Same key, different fingerprint, so you return 422. Is that the right answer?
Where does the fingerprint get computed relative to request decompression and the body-size limit?

Collapse a redelivered event batch into per-aggregate high-water marks

Easy · hashing · at-least-once · aggregation

You drain a batch of up to 5,000,000 events, each (aggregate_id BIGINT, aggregate_version INT, event_type, payload). The log guarantees order within one aggregate only; the batch merges 64 partitions, and a relay failover has redelivered a range, so an older version for an aggregate can appear after a newer one. Given a map of last_applied_version per aggregate, produce the events worth applying, at most one per (aggregate_id, version), plus the count discarded. Target O(n) time. State the memory for 2,000,000 distinct aggregates and what you do when it does not fit.

Approach

One pass, one hash map from aggregate_id to the highest version kept, and a discard counter. An event whose version is at or below last_applied_version for its aggregate is dropped without further work, which is the whole reason the event carries its version rather than a delta. O(n) expected time, O(d) space in distinct aggregates.
Keep the maximum, never the last occurrence. The redelivered range means the final appearance of an aggregate in the batch can be an older version than one seen earlier in the same batch, so last-wins applies stale state over newer state and the projection regresses with no error anywhere.
Cost the memory instead of calling it large: an 8-byte key plus a 4-byte version is 12 bytes of payload, and an open-addressed table held at a 0.7 load factor costs roughly 17 bytes per entry before per-slot metadata, so 2,000,000 aggregates is tens of megabytes in a native layout and several times that in a runtime that boxes both key and value.
If the distinct set exceeds memory, partition on hash(aggregate_id) mod P and reduce each partition independently. Every event for one aggregate hashes to the same partition, so the per-partition result is exact and the merge is concatenation rather than a second reduction.

Follow-up

The payload is a patch rather than a snapshot, so applying only the highest version loses the intermediate changes. What changes in your reduction?
How do you detect that version 7 arrived while version 6 was never delivered, and what should the consumer do about the gap?

Stop tag and share joins from fanning out a page

Medium · join fan-out · semi-joins · many-to-many

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?

Hold a per-tenant active cap against concurrent creates

Hard · write skew · isolation levels · constraints

A tenant on the standard plan may hold at most 50 resources with status='active'. The create handler runs SELECT count(*) FROM resource WHERE tenant_id = $1 AND status = 'active', compares to 50, then inserts. Two creates arrive 3 ms apart on different instances and the tenant lands at 51. Name the anomaly, say whether PostgreSQL 16 READ COMMITTED or REPEATABLE READ prevents it and why, then give an implementation that holds the cap at READ COMMITTED with the exact statements. Finally, say what changes when the cap is 'at most one running export per tenant' on job_run.

Approach

Name it: write skew. The two transactions read an overlapping set and write disjoint rows, so there is no row-level conflict for the engine to detect and each commit is individually legal.
Rule out the levels precisely. READ COMMITTED takes a fresh snapshot per statement and takes no lock on the counted rows, so both see 49. PostgreSQL's REPEATABLE READ is snapshot isolation: it removes non-repeatable reads and phantoms within the snapshot but still admits write skew, because the anomaly is not a re-read of a changed row, it is a read of a set that a concurrent transaction invalidates. Only SERIALIZABLE closes it, by tracking the read dependency and aborting one transaction with SQLSTATE 40001 — a guarantee that exists only if the application re-runs the whole transaction from the read.
Convert the set predicate into a single-row conflict: keep tenant.active_resource_count and run UPDATE tenant SET active_resource_count = active_resource_count + 1 WHERE tenant_id = $1 AND active_resource_count < 50 in the same transaction as the INSERT. Zero affected rows is the cap, returned as 409. The row lock serialises the decision at any isolation level, and contention is bounded to one tenant's row — which is also the fair-scheduling unit, unlike a global counter that would convoy every tenant behind one row.
State the cost you just took on: a counter is a second source of truth that can drift, so every path that changes status must adjust it inside the same transaction, and a periodic reconciliation has to exist, with resource_revision as the authority for what the count should have been.

Follow-up

A resource moves from archived back to active. Which statements change, and what breaks if the counter update and the status change land in different transactions?
The cap becomes plan-dependent and a plan can change mid-month. Where does the number 50 live, and who reads it?

What specific engineering metrics would you use to evaluate performance in a high-load scenario?

Medium · Technical and Domain Expertise

Approach

Frame the answer with the RED and USE views: for each service, Rate (requests or messages per second), Errors (failed share, split by type) and Duration; for each resource, Utilisation, Saturation and Errors (CPU, memory, disk and network I/O, connection pools, thread pools, queue depth).
Report latency as percentiles (p50, p95, p99, max) from histograms, never a mean: one slow dependency moves p99 while the average barely changes, and percentiles from separate hosts cannot be averaged, so aggregate the histogram buckets and then compute the percentile.
Tie the numbers together with Little's law: in-flight work = arrival rate x time in system. If throughput holds while latency climbs, work is queuing somewhere, so check queue depth, pool wait time and GC pauses before adding hardware.
For a data pipeline, such as one feeding plant performance dashboards, add consumer lag, end-to-end freshness (event time to visible time), retry and dead-letter counts, and backlog drain time once load drops.
Anchor each metric to a target: an SLO such as p99 under a stated bound at a stated request rate, an error budget, and a load test that ramps an open-loop (fixed arrival rate) generator to the knee of the latency curve, since closed-loop tools hide stalls (coordinated omission).

Follow-up

p99 latency doubled under load but CPU sits at 40%. Where do you look first, and why?
How would you alert on these metrics without paging someone for every short spike?
How would you find the load level at which the system stops meeting its target?

Explain your former technical experience and demonstrate how it would apply directly to our current infrastruc

Medium · Technical and Domain Expertise

Explain your former technical experience and demonstrate how it would apply directly to our current infrastructure.

Approach

Choose one system you owned and outline it as components: inputs, processing, storage, outputs and who used it. State its scale in concrete numbers (records per day, users, sites, uptime target) and which parts you personally designed or ran.
Spend most of the answer on the decisions: the constraint behind each, the alternative you rejected and the cost you accepted (for example, batch versus streaming ingestion, or a relational store versus a time-series store for sensor-style readings).
Cover failure, not only the happy path: how the system detected a bad input or a dead dependency, how it retried without duplicating data (idempotent writes keyed on a message id), and how it recovered and back-filled after an outage.
Map that experience to what candidates report about this role, which involves monitoring plant performance and supporting field operations: store-and-forward buffering for sites with poor connectivity, data validation before readings reach dashboards, and tools field engineers can use without a developer present.
Do not claim to know the employer's systems. Ask what the current stack and data sources are, then say which part of your experience transfers directly and which you would need to learn.

Follow-up

If the link between a remote site and the central system dropped for a day, how would your design behave, and what would the users see?
What would you change in that system if you rebuilt it today?
What was the hardest production incident on that system, and what did you change afterwards?

Listing latency scales with page size, not with filters

Easy · n+1 · latency · query batching

The tenant listing endpoint reads resource filtered by tenant_id and status, ordered by updated_at DESC, and returns each row plus the owner's display name from app_user and the actor of that resource's latest resource_revision. p99 is 55 ms at 10 rows per page and 1.4 s at 200. Database telemetry shows 401 statements per request, each under 1 ms, and nothing in the slow-query log. Diagnose the cause and give the fix, stating the statement count per request and the p99 you expect afterwards.

Approach

Read the counters before forming a theory. 401 statements for 200 rows is one driver query plus two per row, and sub-millisecond execution with an empty slow-query log rules out a bad plan. The time is round trips, which is why it is invisible in every per-query metric and scales with rows returned rather than with filter selectivity.
Name the two per-row statements from their normalised text: a single-row app_user lookup by user_id, and a resource_revision lookup by resource_id ordered by version DESC LIMIT 1. Confirm by dropping those two response fields and watching the statement count fall to one. That locates the calls in the serialisation layer, not the repository.
Check that the arithmetic accounts for the whole gap. Measure one round trip to the replica in isolation; 400 trips at roughly 3 ms of network plus 0.2 ms of execution is about 1.3 s on top of a 55 ms baseline, which matches. If the multiplication had fallen short, the N+1 would only be part of the story and you would keep looking.
Batch both lookups. Collect owner_user_ids and resource_ids from the driver query, then issue WHERE tenant_id = $1 AND user_id = ANY($2) for the users, and PostgreSQL's SELECT DISTINCT ON (resource_id) ... WHERE resource_id = ANY($2) ORDER BY resource_id, version DESC for the latest revision, which the UNIQUE (resource_id, version) index serves directly. On an engine without DISTINCT ON, use a lateral join or a row_number window. Three statements per request at any page size.

Follow-up

The page size is capped at 200 today. What breaks first if it is raised to 2,000, and is it still this bug?
How do you stop the next N+1 from reaching production, given that no individual query is slow and the endpoint's tests pass?

Suggested practice checklist

The plan starts by mapping the five reported rounds and your resume, then gives each reported theme its own day: brittle failures, safety stories, leadership and deadlines, the two technical questions, and motivation. It ends with a full mock run.

Day 1 — Map the loop and audit your resume

Write the five reported stages in order: Application Review, Virtual Screening, Recorded Video Assessment, In-Person Technical Panel, Final Offer Stage. Next to each, note what you would need to show.
Compare your resume with the reported requirements (modern languages, system design, collaborative development) and the extras (industrial systems, analytics tooling, remote or distributed systems). Mark gaps and strengths.
For every project on the resume, write one obstacle and one trade-off.

Deliverable: A one-page loop map plus a resume audit listing, for each project, its obstacle and trade-off.

Day 2 — Risk management and brittle failures

Choose a system that failed brittly and write the failure as a short STAR story.
Build a failure-mode table for that system: failure, likelihood, impact, detection, control. Include a dropped-connection row.
Rehearse the answer out loud and cut it to the decision you made and the result.

Deliverable: A failure-mode table for one real system and a rehearsed STAR answer on risk management and brittle failures.

Day 3 — Safety stories and your safety record

Write the unsafe-work story: hazard, who was at risk, your decision, the pushback, the resolution and the process change.
Write your safety record answer: one factual sentence, your practices, and one near miss or mistake with what changed.
Check every claim against what you could document in a background check.

Deliverable: Two written STAR stories, one on stopping unsafe work and one on your safety record, each with a documented fact base.

Day 4 — Leadership and missed deadlines

Write the deadline answer as diagnosis, options, stakeholder message and follow-up, using a real example.
Write your leadership style in one sentence with two behaviours, then attach a story with a result.
Record both answers on video and review the playback for structure and rambling.

Deliverable: Two recorded answers, one on handling a teammate who cannot finish a task in time and one on your leadership style and motivation.

Day 5 — Metrics under load and applying your experience

Write a metrics sheet for a high-load service: latency percentiles, throughput, error rate, saturation, queue depth, and the action each triggers.
Prepare how a past system of yours would apply to plant or field operations, covering reliability, remote conditions and who uses it.

Deliverable: A one-page metrics sheet with actions, a written account of how your past system translates to operations, and a worked debugging diagnosis.

Day 6 — Motivation, work history and the recorded format

Write the why-this-company answer using only facts you verified, plus a goal that grows out of this role.
Write the present, past, future version of your work history ending on why you fit.
Record both as a video response and fix filler, pacing and missing results.

Deliverable: Two recorded answers, one on why you applied and your goals and one on your work history and fit.

Day 7 — Full mock panel and offer-stage questions

Run a mock panel with a friend: one failure-mode design discussion, then the risk, unsafe-work and deadline questions.
Practise explaining one trade-off to an engineer and to a non-technical manager.
Write your questions about the team, stack and site operations, and list your constraints on location and shift work.

Deliverable: A scored mock-panel notes page with weak answers marked for a second pass, plus your list of questions and constraints.

Safety, risk and leadership stories

Reported topics for this loop are risk management, unsafe-work prevention, safety record, leadership and motivation for applying, with the STAR method recommended for structure. Safety is the thread to prepare most carefully, so rehearse stories in which you made the call, state the result, and keep every detail accurate. The prompts below are the reported ones.

Describe your experience with risk management and handling brittle failures in software systems.

Medium · Technical and Domain Expertise

Approach

Choose a system that was brittle in a specific way: it worked until one dependency slowed, one input was malformed or one link dropped, then failed completely instead of degrading. Name the failure in one sentence and say what you owned.
Show how you assessed the risk: list failure modes (dependency outage, bad input, partial writes, overload, bad config), rank each by likelihood and impact, and say which you chose to address first and why.
Name the controls you added and match each to a failure mode: timeouts, retries with backoff and jitter, circuit breakers, idempotent operations, input validation, graceful degradation, feature flags with staged rollout, and a rehearsed rollback.
Cover detection and verification: alerts on error rate and tail latency, a fault-injection or failure drill to prove the fix, and a postmortem that fed back into a checklist. Report results in your own measured numbers, never invented ones.
If your example involved intermittent connectivity, say so: store-and-forward queues, local buffering and reconciliation on reconnect are directly relevant to the remote environments candidates report.

Follow-up

Which failure mode did you not anticipate, and how did you find out?
How do you decide that a remaining risk is acceptable to ship with?
How did you test the recovery path before it was needed?

Describe a time you stopped unsafe work; what was the resolution and what did you learn?

Medium · Behavioral and Safety Culture

Approach

Open with a real hazard and who could have been harmed: operators, users or colleagues. If you have no physical-site story, use the nearest honest one, such as a release that could have corrupted data or disabled a protection, a skipped review on a safety-relevant change, or a lockout step being bypassed. Do not inflate it.
Make the stop your decision. Say what authority you had, what you actually did (halted the deploy, paused the task, refused to sign off), and how you communicated it: directly, in writing, and to the person who could resolve it.
Cover the pushback. Name who disagreed, what pressure they were under (schedule, cost), and how you handled it with facts instead of rank.
Give the resolution in concrete terms: the hazard was fixed or controlled, work resumed, and what it cost in time. Then state the learning as a change you made afterwards, such as a pre-release gate, a checklist item or a clearer escalation rule.
The story fails if you only reported the problem to someone else, if the hazard was trivial, or if the ending is a lecture with no changed practice.

Follow-up

What did stopping the work cost, and who disagreed with you?
What would you have done if a more senior person had told you to continue?
What did you change so the same hazard would be caught earlier next time?

What is your professional safety record?

Medium · Behavioral and Safety Culture

Approach

Answer the question first in a sentence, factually: for example, no recordable incidents, or one near miss you reported and corrected. Candidates report background and safety checks as part of the process, so state only what you can document.
Support the answer with practice, not slogans: how you report near misses, run blameless postmortems, use review and change-approval gates, and handle access or secrets with care. Name one specific habit you keep under deadline pressure.
Include one real near miss or mistake, what you did immediately, what you told others, and what changed afterwards. A record with one handled incident is more credible than a claim of perfection.
If your work has had no physical-safety exposure, define the record you do have, such as production incidents, security issues and compliance duties, and describe it with the same honesty. Do not borrow someone else's incident.

Follow-up

Tell me about a near miss you reported, and what happened next.
Was there ever an incident where your own change was at fault?
How do you keep safety habits when a deadline is tight?

How would you handle a situation where a subordinate is unable to finish a critical task before a deadline?

Medium · Behavioral and Safety Culture

Approach

Start with diagnosis before action: meet the person privately and find out why the task is slipping. Typical causes are a skill gap, a blocker you can remove, too much on their plate, unclear scope, or something outside work. The right response differs for each.
Protect the outcome with options: cut scope to the must-have part, re-sequence the work, pair them with someone, take a slice yourself, or move the date. Do not remove testing or safety checks to hit the deadline.
Tell the stakeholder early with a plan and a choice, not on the day it slips. Say what is at risk, what the options cost and what you recommend.
Close the loop: check progress at short intervals, and afterwards look at the cause (estimates, onboarding, tooling) so it does not repeat. If you have not managed a direct report, tell the story with a junior colleague you mentored.
The mistakes that sink this answer are silently taking the work over, blaming the person, or hiding the risk until it is too late.

Follow-up

What if the deadline is fixed by a safety or regulatory requirement and nobody else can help?
How would you handle the same person missing a second deadline?
How did you word the warning to your own manager or the stakeholder?

What type of leader are you, and how do you motivate your team?

Medium · Leadership and Situational Judgment

Approach

Describe your style in one sentence anchored by two observable behaviours, for example setting clear context and checkpoints while leaving the how to the owner, or leading through influence when you have no authority. Avoid a list of adjectives.
Follow with one story where the style produced a result: a team or a stakeholder group you led, a decision you shaped, and a measurable change. Say what you did and what others did, so credit is not all yours.
Explain motivation concretely: learn what each person wants, give ownership of a real piece of work, connect tasks to the outcome for the people who use the software, and recognise progress specifically. Reports describe working across departments and translating technical requirements for non-technical stakeholders, so include a cross-team example.
Show range by naming a situation where you changed style, for example more direct guidance for a new joiner and more delegation for an experienced engineer.
Avoid fashionable labels with no evidence, and avoid describing leadership as being the most senior or most technical person.

Follow-up

Tell me about a time your leadership approach did not work.
How do you lead people who do not report to you?
How would you motivate a teammate who is exhausted and behind?

Why did you apply to Barrick Mining, and what are your long-term professional goals?

Medium · Leadership and Situational Judgment

Approach

Build the answer in three parts: why this kind of work, why this company, why you. For the work, use what candidates report about the role: software for plant performance, resource management and safety tools used by field teams.
Do real research before the interview and use only what you verify: public information on the company's operations, and its DNA-Core Values, which candidates report coming up. Name one value and tie it to something you did. Do not guess at facts about the company.
Link your background to the reported requirements: modern programming languages, system design, collaborative development, and any of the listed extras (industrial settings, data analytics, remote or distributed systems) you truly have.
State long-term goals that grow out of this job, for example becoming the person who designs reliable software for operational environments or leading a team that builds it. A goal that points away from the role weakens the answer.
Avoid praise with no specifics, and avoid sounding as if the company is a stepping stone.

Follow-up

What do you already know about the company's operations?
Which of the company's values do you find hardest to live up to, and why?
Would shift-based work or time at a remote site be a problem for you?

Tell us about yourself and your work history; why are you the right candidate for this role?

Medium · Leadership and Situational Judgment

Approach

Use a present, past, future shape: where you work now and what you own, two or three earlier steps chosen because they relate to this role, and why this job is the next step.
Pick milestones that match the reported requirements: a system you designed, a project you built collaboratively, any exposure to data pipelines, operational tooling or remote and distributed systems. Cut the rest of your history to a line.
End with the second half of the question: why you are the right candidate. Name three matches between what the role involves and what you have done, each with a concrete example, such as reliability work, working with non-engineers, or taking a project from requirements to deployment.
Rehearse it out loud until it flows without notes. Make sure every project you mention is one you can discuss in depth, including the obstacles, because candidates are advised to expect that.
Avoid reciting the resume in order, and avoid stopping before you say why you fit.

Follow-up

Pick one project you mentioned and tell me about the biggest obstacle in it.
What would you want to learn in this role that you could not learn in your current one?
What is the gap in your background that worries you most for this job?