DMY Engineering Consultants Software Engineer Interview Guide

The 60-second brief

DMY Engineering Consultants is described by candidates as an employer of geotechnical and geological project teams working on infrastructure, with projects that range from highway construction to geotechnical site assessments for public and private sector clients. The Software Engineer title sits inside those teams. The reported duties are planning and coordinating field investigations with GIS and GPS technology, taking samples and logging test borings, writing reports in gINT, and working alongside senior engineers on project deliverables. In practice the job turns what is observed on site into records that other engineers can rely on.

This guide is for candidates applying for the Software Engineer position at DMY Engineering Consultants, including entry-level applicants (candidates report 0-1 years of experience is accepted for staff roles). It covers the five reported questions on GIS/GPS use, test borings, engineering software, field data accuracy and juggling tasks, how to answer each, and a seven-day schedule. It does not cover coding or system design, because none has been reported for this loop.

No round sequence has been reported. Candidates describe questions on technical and domain competency (GIS/GPS, test borings, gINT, accuracy) and on behavioral and situational reasoning, and say the pace varies by office and project urgency.

Focus areas

• Geotechnical engineering
• gINT (geotechnical data software)
• Geotechnical field work planning

Overview

The title says Software Engineer, but the reported work is geotechnical. Candidates describe a role built around field investigation: planning where and how to investigate a site, collecting samples, logging test borings, and turning the results into reports. The tools named are GIS, GPS and gINT, a geotechnical data package. Reported topics include geotechnical field work planning, test borings and boring log documentation. Preparing with algorithm sets or system design practice would miss what has actually been reported.

The reported requirements shape the questions you can expect to answer with evidence. A bachelor's degree in Geology, Civil Engineering or a related science is listed, along with strong oral and written communication, the ability to lift 60 pounds and work on uneven terrain, authorization to work in the United States, and a background check that includes motor vehicle and credit history. Internships or work in a related engineering or geological field are listed as a plus. If your background is software rather than geology, your task is to find the closest real overlap: mapping or location data, structured data entry and validation, report generation, or lab and field coursework.

So preparation is mostly about having specific, true stories ready. For each reported question, pick a project and be able to say what you did yourself, which tool you used, what could have gone wrong with the data, and how you caught it. Rehearse the boring log in detail: what is recorded at each sample, how it ends up in a report, and where transcription errors creep in. Then prepare one prioritization story with a real conflict between deadlines. The rest of this page breaks those pieces into a seven-day schedule.

Where to focus your preparation

No round order, round count or duration has been reported for this company. What candidates describe is a set of question areas, so treat the single entry below as a preparation map and confirm the actual format with your recruiter.

Preparation focus

Candidates do not report a round sequence for this company. The reported material is five questions in two groups: technical competency and domain knowledge (GIS or GPS use for data collection and planning, logging test borings or managing geotechnical investigations, proficiency with tools such as gINT, and accuracy when reporting field data under tight deadlines) and behavioral and situational reasoning (managing multiple tasks and prioritizing). The process is also described as varying with office location and project urgency. Approach it as a conversation grounded in your own projects: for each domain question, prepare one concrete project, the tool you used, your personal contribution, and the checks that kept the data correct. For the prioritization question, prepare a STAR story that names your criteria. Ask your recruiter for the stage order, who you will meet, and whether a field component is involved.

What to demonstrate

• Hands-on familiarity with GIS, GPS and gINT, as the reported tool questions cover
• Whether you can describe boring logging and investigation work from your own experience
• Habits that keep field data accurate when deadlines are tight
• How you rank and sequence competing tasks

How to prepare

• Write one project story per reported technical question, naming the tool, your role and the data you handled
• Memorize the parts of a boring log and be able to walk through preparing one from field notes to final report
• List the concrete checks you apply before field data goes into a report, such as units, depth sequence, coordinates and sample identifiers
• Ask the recruiter about stage order, interviewers, and any field or software component

Where candidates lose points

Preparing only coding and system design practice because the title says Software Engineer

The reported questions are about GIS/GPS use, test borings, gINT and field data accuracy. Spend most of your time on those, and ask your recruiter whether any coding is involved rather than assuming.

Answering the test boring question with 'I helped with logging' and no detail

Describe the sequence you worked: sampling method, what was recorded at each sample, how soil was described, how groundwater was noted, and how the data reached the report. Name the part that was yours.

Listing tools such as gINT, GIS and GPS devices without saying what you did with them

For each tool, give one task with an output: a map you built, a set of locations you staked or collected, a log or report you produced. If you have not used gINT, say so and name the closest tool and what transfers.

Saying you 'double-check everything' when asked how you keep field data accurate under deadlines

Name the specific checks: units, depth intervals that run continuously, coordinate system and location, sample IDs, comparison with field notes, and a second person's review before release.

Giving a prioritization story with no ranking logic, or glossing over the physical side of the job

State your criteria and what you deferred. Separately, be ready to speak honestly about safe work outdoors and on uneven terrain, since the reported requirements include lifting 60 pounds.

Questions & practice

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

1. 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.
2. 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.
3. 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.
4. 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?

Track a rolling failure rate per destination for circuit decisions

Easy · sliding window · ring buffer · circuit breaker

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

1. 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.
2. 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.
3. 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.
4. 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?

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

1. 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.
2. 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.
3. 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.
4. 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?

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

1. 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.
2. 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.
3. 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.
4. 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?

Find version gaps and relay lag with window functions

Medium · window functions · event tables · frame semantics

outbox_event holds event_id, aggregate_type, aggregate_id, aggregate_version, event_type, payload, status ('pending','published','dead'), attempts, created_at, published_at. A projection is missing rows and you must decide whether the relay skipped events or the consumer dropped them. Write three queries over the last seven days: one listing every aggregate_id whose published aggregate_version sequence has a hole, one giving per-day counts with a running total, and one returning the newest published event per aggregate. For each, say where the window function is evaluated relative to WHERE and LIMIT. PostgreSQL 16.

Approach

1. Gaps: compute lead(aggregate_version) OVER (PARTITION BY aggregate_id ORDER BY aggregate_version) in a subquery, then filter next_version <> aggregate_version + 1 in the outer query. Window functions are evaluated after WHERE, GROUP BY and HAVING and before the outer ORDER BY and LIMIT, so the predicate cannot sit in the same WHERE clause and PostgreSQL 16 has no QUALIFY.
2. Say what the seven-day filter does to the answer: it truncates every partition, so the first row per aggregate has no predecessor inside the window and a hole spanning the boundary is invisible. Widen the window, or join to resource.version as the authority for the true maximum.
3. Running total: SELECT date_trunc('day', created_at) AS d, count() AS n, sum(count()) OVER (ORDER BY date_trunc('day', created_at) ROWS UNBOUNDED PRECEDING). An aggregate inside a window call is legal because grouping runs before windowing. The grouping key is unique per row here so ROWS and RANGE agree, but write the frame anyway — over ungrouped rows with tied timestamps the default RANGE frame pulls in every peer row and the total jumps.
4. Newest per aggregate: DISTINCT ON (aggregate_id) ... ORDER BY aggregate_id, aggregate_version DESC is the cheap PostgreSQL-only form when an index matches that order; row_number() OVER (PARTITION BY aggregate_id ORDER BY aggregate_version DESC) = 1 is the portable form and needs a subquery for the same evaluation-order reason as the gap query.

Follow-up

• Relay failover redelivers events. Does a duplicate break the gap query, and how would you detect one from this table alone?
• Turn the gap check into a continuous monitor rather than a query someone runs after an incident. What does it watch?

How have you utilized GIS or GPS tools to streamline data collection or project planning?

Medium · Technical Competency and Domain Knowledg

Approach

1. Anchor the answer on one project. Name the step GIS or GPS replaced, for example hand-plotting boring locations on paper plans, and the output it produced: a point layer of planned borings with IDs, offsets and stations that the field crew loaded onto handheld units.
2. Show that you handle coordinates correctly. GPS reports latitude/longitude on WGS84 (EPSG:4326), while project drawings usually use a projected system such as State Plane or UTM. I reproject (pyproj, GeoPandas, ArcGIS or QGIS) before I measure distances or buffers, because degrees are not a unit of length.
3. Explain the planning operations. Buffer utilities and rights-of-way to set clearance, clip candidate locations to the site boundary, then export waypoints as GPX, KML or shapefile. Back in the office, join the surveyed as-drilled points to the boring records by boring ID so the planned and actual locations can be compared.
4. Be specific about accuracy. A handheld receiver is typically good to a few metres, while RTK or differentially corrected GPS gets to centimetre level. GPS heights are ellipsoidal, so getting an elevation for a boring log needs a geoid model. Say which level your work needed and how you checked it.
5. Quantify the result: fewer relocation trips, a shorter mobilisation plan, or no more transcription errors between field coordinates and the report. Also name the edge case you ran into, such as a datum mismatch (NAD83 vs WGS84) or poor signal under tree cover.

Follow-up

• Your plotted borings sit a metre or two off from the surveyor's points. How do you find out whether it is a datum or projection problem rather than a field error?
• How would you design offline field collection so two crews editing the same layer do not overwrite each other when they sync?
• Which part of that workflow would you script, and what would the script take as input and produce as output?

What software tools (such as gINT) are you most proficient with in a professional engineering capacity?

Medium · Technical Competency and Domain Knowledg

Approach

1. Rank your tools honestly in tiers: daily use, project use, familiarity only. For each one, name a concrete artifact you produced with it, such as a set of boring logs, a lab summary table, a site map or a script, rather than listing names.
2. For gINT, show that you understand its structure: borehole and sample data sit in a project database, and the logs and reports are drawn from templates kept in a library. That split is why fixing the data updates every report, while fixing a layout means editing the library, not the data.
3. Connect the tool to software skills. You can read the project data with SQL or pandas, script the import of lab results from CSV instead of retyping them, and check that every sample row points to an existing boring. Mention interchange formats such as AGS or DIGGS if you have used them.
4. If you have not used gINT, say so plainly and bridge from tools you know: other database-backed reporting tools, GIS, Excel or Python data pipelines. Then outline how you would get up to speed, starting from a sample project and the template library.
5. Close with a trade-off you have made, such as editing logs by hand for a one-off job versus building a reusable template or import routine when the same layout repeats across projects.

Follow-up

• A lab sends results in a spreadsheet layout that does not match the project's tables. How would you get them in without retyping?
• What would you keep under version control in a gINT-style setup, and why the templates and not only the data?
• How do you check that a generated boring log matches the field sheet it came from?

How do you ensure accuracy when reporting technical field data under tight project deadlines?

Medium · Technical Competency and Domain Knowledg

Approach

1. Describe validation at the point of entry, not only a review at the end. Depth intervals must be non-negative, with each top above its bottom; sample depths must fall inside the boring's total depth; units must be consistent (feet vs metres); boring IDs must match the plan.
2. Name checks you can automate. Sort strata by top depth and flag any bottom that is greater than the next top (an overlap) or less than it (a gap); the sort makes this O(n log n). Recompute SPT N-values from the raw blow counts (N is the sum of the second and third 6-inch increments) and check refusal entries against the criterion the project uses.
3. Keep a traceable chain: raw field sheet or photo, then database record, then report. Every correction is logged with who made it and why. Never silently overwrite a field value; flag it and resolve it with the person who logged it.
4. Under deadline, triage by consequence. Fully verify the values that drive design decisions first, such as groundwater readings with date and time, refusal depths and strength data. Clearly mark any value that is still waiting on lab confirmation rather than leaving it out or guessing. A second person checks transcribed values against the source.
5. Give one real example: the error you caught or missed, which check would have caught it, and the change you made afterwards, for example adding that check to an import script or a review checklist.

Follow-up

• A report has already gone out and you find a wrong groundwater depth in it. What do you do, in what order?
• If you could automate only three of those checks, which would you pick and why?
• How do you tell a real anomaly in field data, such as an unexpected soft layer, from an entry error?

Edge instances grow 400 MB per hour until the nightly restart

Hard · memory leak · heap analysis · metric cardinality

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

1. 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.
2. 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.
3. 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.
4. 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?

Suggested practice checklist

Seven days built from the five reported questions and the reported tools (GIS, GPS, gINT, boring logs). Each day produces something written, so by day seven you have rehearsed stories rather than general notes.

Day 1 — Map the role against your own experience

• Write two columns: the reported duties (GIS and GPS field planning, sampling and logging test borings, gINT reports, working with senior engineers) and your real evidence for each
• Mark each gap and decide the closest honest substitute, such as coursework, a lab, an internship, or mapping and data validation from other work
• Check the reported requirements (degree, lifting 60 pounds on uneven terrain, work authorization, background check) and note anything to raise with the recruiter

Deliverable: A two-column evidence sheet with gaps marked and a list of recruiter questions.

Day 2 — GIS and GPS story

• Pick one project where location data shaped planning or collection and write out the before and after
• List the practical points you handled: coordinate system, accuracy of the device, field verification of points, and how the data reached the team
• Practice a short spoken answer to the GIS/GPS question without reading it

Deliverable: One written project story and a spoken run of the answer.

Day 3 — Test borings and geotechnical investigations

• Sketch a boring log from memory: header information, depth column, sample type and number, recovery, blow counts, soil description, groundwater, notes
• Write the sequence of an investigation from planning locations to final report, marking which steps you personally performed
• Write your answer to the question about logging test borings or managing investigations, using a real example

Deliverable: A blank boring log layout and one written experience answer.

Day 4 — Tools, with gINT in focus

• List every professional tool you have used and your level for each, with one task and output per tool
• Describe a gINT-style workflow in your own words: enter field data, check it, generate the log or report from a template
• If gINT is new to you, name your closest tool and what carries over, and say so directly in your answer

Deliverable: A tool inventory with an example output for each entry and a prepared answer to the software tools question.

Day 5 — Accuracy under deadline pressure

• Build a pre-release checklist for field data: units, continuous depth intervals, sample IDs, coordinates, comparison with field notes, second reviewer
• Recall one time an error was caught, by you or someone else, and write what the check was and what it prevented
• Practice explaining how you keep these checks when a deadline is tight

Deliverable: A written checklist and one error-caught example that supports your answer on field data accuracy.

Day 6 — Prioritization and reversal stories

• Write the multiple-tasks story in STAR form, with explicit ranking criteria, what you deferred and who you told
• Prepare one story where plans changed because of conditions you found, as a second behavioral example

Deliverable: A STAR outline for the prioritization question and one backup story.

Day 7 — Full run-through and recruiter questions

• Answer all five reported questions aloud in sequence, using your written stories as a reference only
• Tighten any answer where you could not name the tool, your role, or the check you used
• Write the questions you will ask: stage order, interviewers, field versus office split, and equipment or safety requirements

Deliverable: A completed run-through with revised notes and a final list of questions for the recruiter.

Behavioural preparation

The reported behavioral material is small: one situational question about handling several tasks at once and a domain-experience question about test borings and investigations. Candidates also describe the role as one where plans change with site conditions and where accuracy matters under pressure. Prepare stories where you decided something yourself, kept the records correct, and communicated clearly with the people relying on your work.

1. Describe a time a plan had to change because of conditions you found, and how you adjusted.
2. Tell a story about catching an error in data or a report before it was released.
3. Give an example of explaining findings clearly to a senior engineer or client who needed to act on them.

Describe your experience with logging test borings or managing geotechnical investigations.

Medium · Technical Competency and Domain Knowledg

Approach

1. Choose one investigation you can describe in detail and open with its purpose and your exact role: logger, sampler, field coordinator or report writer. Say how many borings or how large a site only if you remember it truthfully. Vague claims such as 'I assisted' lose the answer; say what you did.
2. Walk through the work in order: planning boring locations and depths, clearing the site, sampling method, then the log itself (sample depths, recovery, blow counts if a split-spoon sampler was used, soil description under a classification system such as USCS, groundwater observations, notes on drilling behavior), and finally transfer into gINT or comparable software.
3. Show a judgment call. For example, a soil layer that did not match expectations, poor sample recovery, or a boring stopped early, and what you recorded and who you told. The decision and its reasoning matter more than a list of tasks.
4. Close with the outcome: which report, design question or deliverable used your log, and what a senior engineer or lab result later confirmed or corrected. If your experience is coursework, an internship or a lab, say so plainly and describe the closest real work instead of overstating it.

Follow-up

• What did you do when a layer or groundwater reading did not match what the plan expected?
• How did you check your log against lab results or a reviewer's comments, and what was corrected?
• If you were asked to plan the same investigation again, what would you change?

Tell us about a time you had to manage multiple tasks simultaneously; how did you prioritize your workload?

Medium · Behavioral and Situational Reasoning

Approach

1. Use STAR with a real clash of deadlines: two or three deliverables due close together, such as field data that needed entering, a report that needed finishing and a request from a senior engineer. Keep the situation to a couple of sentences and spend most of the answer on how you decided.
2. State your ranking criteria out loud: hard deadline, who is waiting on the output, whether one task blocks another, and the cost of an error in each. Then show the result of applying them, including what you postponed and whom you told.
3. Include a communication step: you confirmed priorities with the person who owned them, or flagged early that something would slip, rather than silently absorbing it. Tie this to precision, since the reported role depends on accurate field records, and explain that you did not skip checking to save time.
4. Give a checkable result (deliverables sent, nothing returned for errors, a schedule renegotiated) and one thing you would do differently. The mistake that sinks this answer is saying you worked longer hours with no ranking logic.

Follow-up

• What did you drop or delay, and who did you tell?
• What if two senior engineers each said their request was the urgent one?
• How did you make sure the rushed work was still accurate?

Reverse your own decision and price the reversal

Medium · reversibility · measurement · migrations

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

1. 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.
2. 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.
3. 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.
4. 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?