Interview guides · AAK

AAK Software Engineer Interview Guide

AAK Software Engineer interview guide: reported rounds, automation and capital project questions, contractor and start-up stories, and a 7-day prep plan.

The 60-second brief

Candidates describe AAK as a producer of plant-based oils. The Software Engineer opening they report sits at the Richmond, California manufacturing plant, where the oil is processed through filtration, deodorization and bleaching. Inside the company the job is reported to carry the title Project Engineer. The work is not application development: it is writing automation functional descriptions, building process models, specifying flow control, preparing process datasheets, updating P&IDs, and running capital projects for new or modified equipment from concept through commissioning.

This guide is for engineers preparing for the AAK Software Engineer (Project Engineer) loop at the Richmond plant. It walks through the three reported rounds and gives worked guidance for all eleven reported questions: the functional description, the material balance and P&ID check, live-plant data, the energy modeling study, the capital project and stage gate stories, schedule recovery, forecasting, contractor oversight, start-up consulting and the scope conflict. A seven-day plan ties them together. It assumes a project-based engineering background and does not teach coding puzzles, because none of the reported questions are coding problems.

Candidates report three rounds over roughly three to five weeks: a Recruiter Screen, a Technical Deep-Dive, and an On-Site Loop at the Richmond plant that includes a plant tour and meetings with cross-functional stakeholders.

Focus areas

Capital Project Management
Schedule Management
Process Engineering (Manufacturing)

Overview

The reported work is an engineering project role with a software-flavoured title. The deliverables are documents and systems that contractors build from: automation functional descriptions, process datasheets and updated P&IDs, plus the process modeling studies behind them. The process being automated is the processing of plant-based oils through filtration, deodorization and bleaching, so those steps are the physical context for every answer. Candidates also report time on the plant floor during commissioning and start-up, in protective gear.

The eleven reported questions fall into three groups. Four are about process engineering and automation: developing a functional description for new equipment, checking that material balances and P&IDs agree before execution, choosing which data points to prioritize when evaluating process schemes from a live plant, and a process modeling study that found an energy efficiency gain. Four are about capital project execution: the most complex project you managed and how you controlled scope, stage gate methodology, recovering a late project, and forecasting project activity for senior management. Three are about contractors and start-up: oversight of a contractor's process design performance, acting as technical consultant during a stressful start-up, and a scope conflict with an engineering contractor.

Preparation therefore means building a bank of real projects with numbers you can defend: budget, schedule, final variance, energy saved, what you personally wrote or decided. Candidates are told to quantify impact, so every story needs a baseline and a measured result. Alongside the stories, prepare the physical fundamentals the requirements list: heat transfer, fluid transport and reaction chemistry, each tied to a control or design decision you have actually made.

If your background is closer to software or controls than to chemical process engineering, the reported requirements describe a senior, project-based engineering profile with chemical engineering preferred. Work out which of your projects is the nearest equivalent, be exact about your own role in it, and rehearse the gaps in plain language rather than leaving them for the interview to find.

The AAK interview process

Candidates describe three stages, from a recruiter conversation to a technical discussion to an on-site loop at the Richmond plant. Each step below says what the stage covers as candidates report it and how to prepare for it.

Recruiter Screen

Candidates report that the Recruiter Screen digs into your capital project background and asks whether you will do the work hands-on at the Richmond plant. The role is reported as fully on-site, with travel of up to 15% to US and foreign destinations, so expect to confirm both. Treat the call as the first use of your project stories: have a short, numeric account of your largest projects ready, and be clear about location, travel and on-site conditions.

What to demonstrate

Depth of experience in capital project execution, as the stage is reported to focus on it
Willingness to work on site in an active manufacturing plant
Fit with the reported requirements: an engineering degree and a long record of project-based work

How to prepare

Write a one-paragraph summary of each of your three biggest projects with budget, schedule outcome and your specific role.
Be ready to say plainly why you want a plant-based role in oil processing rather than a pure software one.
Confirm your position on full on-site work and on travel up to 15%, and ask the recruiter which people you will meet in later rounds and what protective gear the plant visit needs.

Technical Deep-Dive

Candidates describe the Technical Deep-Dive as a technical round led by engineering leadership that covers modeling, automation and stage gates. The reported question types in this area include writing an automation functional description for new equipment, proving material balances and P&IDs are synchronized, choosing data from a live plant to evaluate process schemes, and telling the story of an energy efficiency study. Prepare to work from documents you have actually produced.

What to demonstrate

Ability to turn a physical process need into a precise automation functional description
Working knowledge of material balances, P&IDs and process datasheets
How you use live operating data to model a process and find an improvement
Command of stage gate methodology as used on capital projects

How to prepare

Outline a functional description for a piece of equipment you know well: scope, operating modes, control loops, interlocks, alarms, start-up and shutdown sequences, tied to P&ID tags.
Write a checklist for confirming a material balance, the P&IDs and the functional description describe the same plant before execution begins.
Prepare the energy efficiency study as a story with a baseline from operating data, a model, a recommendation and a measured result.
Be able to walk a project through your stage gate phases and say what each gate decided.

On-Site Loop

Candidates report an on-site loop at the Richmond facility with meetings across functions, including plant managers, operations leads and safety officers, plus a plant tour. Safety awareness is reported to come up around the tour. The behavioral questions on scope control, schedule recovery, contractor oversight and start-up consulting are the natural material, and safety should run through all of it.

What to demonstrate

Communication across operations, safety and management audiences
Track record on scope, schedule and contractor performance
Safety awareness and comfort on an active manufacturing floor, which candidates report coming up on the tour

How to prepare

Rehearse the capital project story, the late-project recovery story and the contractor conflict story, each with one concrete safety or compliance action you took.
Practise explaining a process modeling study to someone who is not a process specialist, in a few clear sentences with the number that matters first.
Prepare for the physical visit: ask what protective equipment is provided, arrive able to walk the plant in the required clothing, and have questions ready about the process areas you will see.

Where candidates lose points

Describing the automation functional description like a generic software spec

Structure it around the equipment: scope, operating modes, control loops, interlocks, alarms and start-up and shutdown sequences, each tied to P&ID tags. Say how you check it against the material balance.

Telling capital project stories with no budget, schedule or variance figures

The reported question asks for the budget directly. Open each project story with approved budget, schedule, final variance and your role, and give the energy saved or payback for any modeling study.

Using stage gate as vocabulary without being able to say what a gate decided

Prepare one project walked through its phases, with the deliverables presented at each gate, who approved, and one case where a gate sent work back.

Claiming software strength but being unable to link heat transfer or fluid transport to a control decision

For each fundamentals topic the requirements list, prepare one relation and one control or design consequence from your own work, such as how pump and valve behaviour affects a flow loop.

Treating safety as an afterthought in contractor and start-up stories

Reports describe safety officers and a plant tour in the on-site stage. Put one concrete safety or compliance action into every story, and describe your part with I, not we.

Questions & practice

Merge partitioned event streams into one ordered feed with bounded lateness

Hard · k-way merge · watermarks · out-of-order streams

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?

Identify the heaviest tenants in a five-minute window under memory pressure

Medium · top-k · heavy hitters · streaming

The edge service handles about 3,000 requests per second across roughly 50,000 tenants, peaking near 9,000. Expose the 50 heaviest tenants by request count over the trailing five minutes so limits can be tightened before one tenant's backfill starves the fleet. You may not retain five minutes of raw records. Give the exact solution and its memory, then the bounded-memory approximation with its error stated as a formula, and say which you would ship and at what tenant cardinality that choice changes.

Approach

Do the exact version first, because it is affordable at this cardinality: a ring of 300 one-second counters per tenant, advanced lazily, is 1,200 bytes of counters per tenant and roughly 60 to 90 MB for 50,000 tenants with overhead. Carry a running total and subtract the bucket you overwrite so a window read is O(1) rather than 300 adds.
Extract the top 50 with a size-k min-heap over the tenant sums: O(d log k) for d tenants, against O(d log d) to sort them all. Maintaining the heap continuously instead of on query requires a tenant-to-heap-index map, because incrementing a count already inside the heap means sifting from a known position, and without that map you rebuild the heap on every request.
State the approximation precisely rather than gesturing at sketches. Misra-Gries with m counters retains every item whose true count exceeds N/(m+1), and each retained count underestimates the truth by at most N/(m+1). With m = 1,000 and N = 900,000 requests in the window the error is roughly 900 requests, which is fine for spotting a tenant sending 50,000 and useless for ranking two tenants 200 apart.
Say what breaks when the window slides: Misra-Gries and Space-Saving are insert-only and cannot be decremented as records age out. The workable construction is one summary per sub-window, say ten seconds, with 30 summaries merged at query time, and the merged error is the sum of the per-summary errors, so the bound degrades linearly in the number of sub-windows.

Follow-up

The heaviest tenant is heavy because of one export job rather than user traffic. Should the limiter treat those as the same tenant?
Two tenants sit tied at the boundary of the top 50. Does your answer flap, and does the flapping matter?

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?

Write the update path that detects a concurrent edit

Medium · optimistic concurrency · transactional outbox · audit history

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

Approach

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

Follow-up

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

Keep soft-deleted accounts from blocking re-registration

Easy · soft delete · unique constraints · engine differences

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

Approach

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

Follow-up

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

Walk me through how you develop an automation functional description for a new piece of processing equipment.

Medium · Process Engineering & Automation

Approach

Start from the process design basis rather than the PLC: the P&ID, process datasheets and material balance give the equipment list, every tagged instrument and valve, the I/O count and the normal, minimum and maximum operating ranges the logic has to cover.
Use ISA-88 for the structure: control modules (each valve, motor and loop) and equipment modules. Keep the vocabularies apart. ISA-88 procedural states are Idle, Running, Complete, Holding/Held, Pausing/Paused, Restarting, Stopping/Stopped and Aborting/Aborted. Start-up, clean (CIP) and shutdown are plant operating phases written as numbered sequences with transition conditions, and modes (auto, manual, maintenance) are separate again. Continuous deodorizers often borrow the module structure without the batch state model, so say which parts you adopt.
For each control loop state the process variable, setpoint source, controller type (PID, cascade or ratio), final control element and its fail position; for a bleaching vessel under vacuum, that covers temperature, vacuum level and slurry dosing.
List alarms with setpoints and priorities (ISA-18.2 style), permissives and interlocks, each traced back to a HAZOP action; keep safety instrumented functions in a separate IEC 61511 specification instead of the basic control system description.
Cover the abnormal cases explicitly (power loss and restart, loss of communications, a bad-quality transmitter signal, an operator abort mid-sequence), get operations and maintenance to review it, and reuse each section as a FAT/SAT test script under revision control.

Follow-up

How do you specify what the equipment does on restart after a power failure in the middle of a sequence?
Where do you draw the line between a control-system interlock and a safety instrumented function?
How do you turn the functional description into factory and site acceptance test procedures?

How do you verify that your material balances and P&IDs are completely synchronized before a project moves to

Medium · Process Engineering & Automation

How do you verify that your material balances and P&IDs are completely synchronized before a project moves to the execution phase?

Approach

Build a stream-by-stream cross-reference: every stream number on the PFD and heat and material balance maps to P&ID line numbers and the line list, and flow, temperature, pressure, phase and composition agree for each case (design, normal, minimum or turndown).
Check that the balance itself closes: mass in equals mass out plus accumulation for every unit and overall, component balances close, and recycle and utility streams (steam, cooling water, vacuum) are included, within an agreed tolerance.
Check the hardware against the balance: instrument ranges cover the normal and extreme flows, control valves are sized (Cv) at both minimum and maximum flow, line sizes give acceptable velocity and pressure drop, pumps have NPSH available above NPSH required, and relief cases match the scenarios.
Lock the governance: both documents at the same revision and frozen before the gate, a formal multidisciplinary P&ID review and HAZOP on the frozen set, a field walkdown where tie-ins meet existing equipment, and management of change for anything after the freeze.
Name the trade-off: a hard freeze slows late improvements, but every unsynchronised change found during construction costs far more in rework than the same change on paper.

Follow-up

The HAZOP adds a bypass line after the balance is frozen. What do you re-check, and who signs it off?
How do you handle tie-ins where the as-built P&IDs of the existing plant are out of date?
What closure tolerance do you accept on the material balance, and why?

What data points do you prioritize when evaluating process schemes from a live manufacturing environment?

Medium · Process Engineering & Automation

Approach

Start from the decision the evaluation supports (a throughput bottleneck, energy use, yield loss or quality), then choose the KPIs: throughput, yield or loss per tonne, specific energy (steam or kWh per tonne), downtime and, for edible oils, lab quality results such as free fatty acids, colour and peroxide value.
Pull the right sources together: historian tags (flows, temperatures, pressures, valve positions, controller outputs and modes), lab or LIMS results, batch records and maintenance logs, aligned on time and allowing for the lag between sampling and lab results.
Validate the data before trusting it: flat-lined or stuck tags, historian compression, calibration dates, and a mass balance check across the unit to catch a meter that disagrees with the others; filter to steady-state periods.
Look for the bottleneck signatures: control valves pinned near fully open, loops left in manual, high variability in the process variable, frequent alarms, and drift from design, such as a falling exchanger U-value from Q = U·A·ΔT_lm, which points to fouling.
Normalise by rate and feedstock before comparing regimes, treat correlations in historical data as hypotheses, and confirm the key one with a planned plant trial before committing capital.

Follow-up

How do you tell an instrument fault from a real change in the process?
How would you detect heat exchanger fouling from historian data alone?
What period of operating data would you treat as representative, and why?

Describe a situation where a project was falling behind schedule. What specific actions did you take to bring

Medium · Capital Project Execution

Describe a situation where a project was falling behind schedule. What specific actions did you take to bring it back on track?

Approach

Quantify the slip first: compare progress against the baseline schedule, find which critical-path activities moved and by how much, check total float on the near-critical paths, and give the schedule performance index (SPI = EV / PV) if earned value was tracked.
Name the root cause specifically: late long-lead equipment, late engineering deliverables, contractor productivity, a permit, or a fixed shutdown window that turned a small delay into a large one.
Explain the recovery options and why you chose one: crashing (overtime, extra crews; pick the critical activity with the lowest cost per day gained, because crashing a non-critical activity buys nothing) versus fast-tracking (overlapping activities, which adds rework risk), plus re-sequencing, prefabrication or pre-commissioning before the outage, or moving non-essential scope to a later phase.
Describe the execution: an approved recovery schedule, a short look-ahead reviewed with contractors, tracking against it, and watching for a new critical path emerging once the original one is shortened.
Close with numbers in your own voice: the days lost, the days recovered, what the recovery cost, who approved it, and what you changed in planning afterwards.

Follow-up

Why did you choose crashing over fast-tracking, or the reverse, in that case?
What did the recovery cost, and how was that spend approved?
Which activity became critical once the original delay was recovered?

How do you forecast capital project activities and present those financial reports to senior management?

Medium · Capital Project Execution

Approach

Build the forecast from the approved budget broken down by WBS or cost code, including contingency, and track four numbers per line: budget, commitments (POs and contracts), actuals, and accruals for work done but not yet invoiced.
Forecast at completion explicitly: EAC = actual cost + estimate to complete; with earned value, CPI = EV / AC, and EAC = BAC / CPI if the current cost trend is expected to continue, or AC + (BAC - EV) if the variance was a one-off.
Phase the spend over time as a monthly cash-flow S-curve for capital budgeting, and keep a risk register with quantified exposures and a change log of pending and approved change orders so contingency drawdown is traceable.
Present one page to senior management: budget, committed, spent, EAC, variance and its cause, contingency remaining, top risks with mitigations, and the decisions needed, on a fixed cadence so a variance is raised early instead of at closeout.
Cover the edge cases: escalation and currency on imported equipment, unpriced change orders, and reconciling your forecast with the finance team's accruals so the two reports never show different numbers.

Follow-up

When do you release contingency, and who approves it?
Your EAC now exceeds the approved budget. What do you present, and what do you ask for?
How do you reconcile your project forecast with finance's month-end accruals?

p99 jumped on one listing filter while p50 stayed flat

Medium · query planning · composite index · percentiles

After a release that added an owner_user_id filter to the resource listing, p99 rose from 90 ms to 1.9 s while p50 stayed at 40 ms. Traffic and row counts are unchanged. resource carries the index (tenant_id, status, updated_at DESC, resource_id DESC). The new query filters tenant_id and owner_user_id, orders by updated_at DESC, resource_id DESC, and takes 20 rows. On PostgreSQL, explain the shape of the regression, prove it from a query plan, and give the index you would add.

Approach

Start from the shape. A flat p50 with a moved p99 means a subset of requests changed cost, not all of them, so the first job is naming the subset. Bucket the endpoint's latency by the tenant's row count; the natural hypothesis is that large tenants are a small share of requests and all of the tail.
Get the plan for the new query on a large tenant with EXPLAIN (ANALYZE, BUFFERS). Expect an index scan over the tenant's range, a filter discarding most of it, then a Sort feeding the Limit, possibly reporting Sort Method: external merge Disk. Read actual rows on the scan node, not estimated.
Explain why the existing index cannot serve it. A composite B-tree is seekable only as a left prefix, and with no equality predicate on status the scan cannot treat updated_at as an ordering, because rows in the tenant's range are ordered by status first. Everything matching must be read and sorted before LIMIT 20 can apply, so a tenant with 400,000 rows pays 400,000 rows to return 20.
Add (tenant_id, owner_user_id, updated_at DESC, resource_id DESC). Equality on the first two columns leaves the index ordered by updated_at within that pair, so the plan becomes an index scan that stops after 20 rows with no Sort node. PostgreSQL can scan a B-tree backwards, so the DESC markers matter only if the two sort columns ever disagree in direction; keeping them explicit documents the order the keyset cursor depends on.

Follow-up

The endpoint paginates with OFFSET. What does page 500 cost with your index, and what does the keyset version cost?
How would you have caught this before release, given that a 10,000-row seed database produces the same plan shape at an unnoticeable cost?

Suggested practice checklist

The plan starts with an inventory of your real projects, then takes the reported question groups in order: automation documents, process modeling, fundamentals, capital project execution, and contractors and start-up. The last day is a full run-through for the on-site stage.

Day 1 — Map the loop and inventory your projects

Write the three reported rounds (Recruiter Screen, Technical Deep-Dive, On-Site Loop) and next to each the question types it is likely to carry.
List your three or four strongest projects with budget, schedule outcome, your role, and one measurable result each.
Mark the gaps: any project area you cannot yet back with a number.

Deliverable: A one-page loop map and a project table with budget, schedule outcome, role and result for each project, gaps marked.

Day 2 — Automation functional descriptions and document consistency

Outline a functional description for a piece of equipment you know, with operating modes, control loops, interlocks, alarms and start-up and shutdown sequences.
Write a checklist for confirming the material balance, the P&IDs and the functional description agree before execution.
Say your answers to the functional description question and the synchronization question out loud.

Deliverable: A functional description outline, a document consistency checklist and two spoken answers.

Day 3 — Process modeling and live operating data

List the data points you would pull to evaluate a process scheme in a live plant (flows, temperatures, pressures, energy use, quality results, downtime) and why each one comes first or later.
Write the energy efficiency study as a story card: baseline, model, assumptions, opportunity, payback and measured result.

Deliverable: A prioritized data list with reasons, and one story card with before-and-after figures.

Day 4 — Engineering fundamentals tied to control decisions

Heat transfer and fluid transport: write Q = U·A·ΔT_lm for heat recovery between deodorizer feed and product, and the pump curve against system curve, NPSH and valve authority for flow loops on viscous oil, each with its control consequence.
Reaction chemistry: note bleaching-earth adsorption and the oxidation sensitivity of the vacuum and temperature window. General edible-oil practice strips in the deodorizer at high vacuum and roughly 230-260 °C, trading free fatty acid removal against trans-fat and 3-MCPD or glycidyl ester formation.
Prepare how you would explain the way fluid transport shapes flow control logic, using an example from your own work.
Note the filtration, deodorization and bleaching steps in the oil process and one thing each would need from an automation system.

Deliverable: A one-page fundamentals sheet with a control consequence for each topic, plus an outline for a two-minute spoken explanation of how fluid transport shapes your flow control logic.

Day 5 — Capital project execution

Write your stage gate walkthrough: phases, the deliverables at each gate, approvers, and one gate that sent work back.
Prepare the complex project story and the late project recovery story with budget, variance and the specific actions you took.
Build a sample forecast table (committed, spent, forecast to complete, variance) and rehearse presenting it to senior management.

Deliverable: A stage gate walkthrough, two project stories and a forecast table, each spoken aloud once.

Day 6 — Contractors, commissioning and conflict

Write your contractor oversight method: specifications, review milestones, hold points, comment log and nonconformance steps.
Prepare the start-up consulting story, including your sequence for unexpected process questions.
Prepare the scope conflict story with the documents you used and how it was resolved.

Deliverable: An oversight checklist and two story cards (start-up and scope conflict), spoken aloud.

Day 7 — Full run-through and on-site readiness

Run the recruiter, technical and on-site material back to back out loud and cut any story that lacks a number.
Add one concrete safety action to each story.
Write your questions for the recruiter about the plant visit, protective gear and who you will meet.

Deliverable: A final story list with safety lines added and a list of questions for the recruiter.

Project, contractor and start-up stories

The reported behavioral questions ask for projects you ran: a process modeling study, a capital project, stage gate experience, contractor oversight, a plant start-up and a scope conflict. Each story needs your own actions, numbers for budget, schedule or energy, and a measured outcome. Safety belongs in each one, because the loop is reported to include safety officers and a plant tour.

Describe a time you performed a process modeling study that successfully identified a major energy efficiency

Medium · Process Engineering & Automation

Describe a time you performed a process modeling study that successfully identified a major energy efficiency opportunity.

Approach

Choose a study where you built or ran the model yourself, and name the process step it covered. Open with the baseline taken from operating data, such as energy used per unit of throughput across a representative range of operation, not a single good week.
Show the modeling: the heat and material balance you set up, the assumptions you made (heat capacities, fouling, equipment efficiency), and how you reconciled the model to plant measurements before trusting any prediction from it.
State the opportunity in numbers: energy saved per unit and per year, capital or control changes needed, and simple payback. Name the one assumption the result is most sensitive to.
Close the loop: how the change was implemented, what the measured result was against the prediction, and any shortfall. Say which product quality or safety limits the change had to respect.

Follow-up

How did you validate the model against plant data before you relied on it?
Which assumption moved the result most, and how did you test it?
How did you get operations to adopt the change?

Tell me about the most complex capital investment project you have managed. What was the budget, and how did y

Medium · Capital Project Execution

Tell me about the most complex capital investment project you have managed. What was the budget, and how did you control scope deviation?

Approach

Pick one project and answer the budget part in your first sentence: approved budget, schedule, what was built or modified, and your role (premise, budget, schedule, contractor oversight, commissioning). Use only figures you can defend.
Explain how scope was fixed at the start: a written project premise or scope definition, and a change-control step where every request had its cost and schedule impact written down and an approver named.
Give one or two real scope pressures, how you assessed each, and whether you accepted, rejected or deferred it. Mention contingency and how it was spent.
Report the outcome against baseline: final cost versus approved, schedule variance, whether performance targets were met, and the lesson you took into the next project.

Follow-up

How much contingency did the project carry and how did you spend it?
Which change request did you reject, and who pushed back?
What would you set differently at the start next time?

Explain your experience with stage gate methodology. How do you ensure stakeholders are aligned before passing

Medium · Capital Project Execution

Explain your experience with stage gate methodology. How do you ensure stakeholders are aligned before passing a gate?

Approach

Define stage gate in your own words: phased project work with approval points between phases, where each gate ends in a decision to proceed, hold, rework or stop. Then name the gate structure you actually worked under and map its phases to the standard idea.
List what you prepared for a gate: scope definition, an estimate whose accuracy fits the phase, schedule, risk list, safety review and operations input, so the decision rested on documents rather than opinion.
Describe how you secured alignment before the meeting: pre-reads, one-to-one conversations with each approver, written entry criteria, an open-items list with owners, and a decision log afterwards.
Include one case where the process mattered: a gate that sent work back because the estimate or operations input was incomplete, and what that saved.

Follow-up

Tell me about a project that failed a gate.
What do you do when an approver signs off and later reopens scope?
How does estimate accuracy change between early and late phases?

How do you provide oversight to ensure a contractor’s process design performance meets your exact specificatio

Medium · Cross-Functional Leadership & Start-Up

How do you provide oversight to ensure a contractor’s process design performance meets your exact specifications?

Approach

Start with the standard you oversee against: functional descriptions, process datasheets and P&IDs written precisely enough to be checked, with performance criteria stated as measurable values such as capacity or response, so oversight is comparison against a written reference.
Describe the mechanism: document review milestones with a comment-and-response log, hold and witness points for inspections and tests, regular progress meetings with a look-ahead, and a rule that every P&ID change flows into the functional description.
Show how you catch drift early: compare submittals to datasheets, track open comments, and check that vendor assumptions match the design basis.
Handle a miss with a short STAR example: the gap you found, a written nonconformance, the corrective action agreed, who bore the cost or time, and your own action throughout. Add contractor safety requirements as part of oversight.

Follow-up

The contractor says the deliverable meets the intent but not the specification. What do you do?
How do you oversee a contractor whose people know a piece of equipment better than you do?
Schedule pressure is high and the contractor asks you to accept a deviation. How do you decide?

Tell me about a time you acted as a technical consultant during a stressful plant start-up. How did you handle

Medium · Cross-Functional Leadership & Start-Up

Tell me about a time you acted as a technical consultant during a stressful plant start-up. How did you handle unexpected process questions?

Approach

Set the scene in two sentences: the system, the start-up stage (pre-start checks, first run, ramp-up) and why your role as the technical reference mattered.
Show preparation that reduced surprises: a walk-down comparing P&IDs with installed equipment, loop checks, a written start-up procedure and a list of known risks agreed with operations and safety.
Describe your sequence for an unexpected process question: secure a safe state first, gather live trends and field readings, compare with the design basis in the datasheet and functional description, decide, and tell everyone the decision. If you did not know the answer, say how you found it and who you brought in.
Give one specific question and the reasoning to its resolution, then close with the documentation you updated afterwards: procedure, functional description, P&ID or punch list.

Follow-up

Describe a moment you stopped or held a start-up step, and why.
What did you do when you genuinely did not know the answer?
What changed in the documents after the start-up?

Describe a conflict you had with an engineering contractor regarding project scope. How was it resolved?

Medium · Cross-Functional Leadership & Start-Up

Approach

Describe the disagreement factually: what the contract and specification said, what the contractor claimed, and the cost and schedule at stake. Keep blame out of the telling.
Show that you went to the documents first: scope definition, datasheets, change log and meeting records, separating a clarification of existing scope from a genuine change.
Walk through the resolution path in order: direct conversation, a written position, a joint review of the facts, the quantified impact, and escalation only when those steps failed. State the outcome, such as a contractor absorbing the work or a change order approved with an offsetting adjustment.
Finish with the relationship and the lesson: how the next phase ran, and what you changed in how scope is written or reviewed. The stories that sink are the ones that cast the contractor as the villain or leave your own actions out.

Follow-up

What if the contractor had turned out to be right?
How did you document the outcome?
When did you involve your own management, and why then?