Why Offshoring Engineering Design Stops at Responsible Charge

The email went out on a Tuesday. Production drafting and a chunk of the calculation work move to the firm’s overseas office next quarter, and the local teams “transition to review and client-facing roles”. The part nobody said in the town hall is the part that matters to you personally: your seal still goes on the drawings.

That deserves a straighter answer than the thread gives. Offshoring engineering design in AEC runs into a constraint software offshoring never meets: responsible charge cannot be relocated. The engineer of record has to exercise real professional control over work they did not personally produce, and be able to demonstrate it afterwards. Every offshore design arrangement that functions is built backwards from that requirement. The ones that fail were built forwards from a rate card.

So the useful question is not whether overseas engineers are capable. Many are excellent. It is which work can be produced remotely and still be defensible under a seal, and what that costs.

Responsible charge is the part that cannot be relocated

The NSPE Code of Ethics is short and explicit. Section II.2.b: “Engineers shall not affix their signatures to any plans or documents dealing with subject matter in which they lack competence, nor to any plan or document not prepared under their direction and control.” Two prohibitions in one sentence, and the second is the offshoring constraint in nine words. II.2.a adds that engineers “shall undertake assignments only when qualified by education or experience in the specific technical fields involved”. II.2.c permits an engineer to coordinate and seal a whole project, “provided that each technical segment is signed and sealed only by the qualified engineers who prepared the segment”.

Read II.2.c carefully, because people reach for it and it does not say what they want. It distributes responsibility among engineers who each carry their own. It does not let one engineer absorb it for a production shop.

The clearest published account of what supervision has to involve is Engineers Canada’s national guideline on direct supervision. The line that settles most arguments: “supervising engineers who conduct only a final review of documents, and who are unaware of the work prior to those documents arriving on their desk, have not fulfilled their supervisory role.” It then describes how involvement is shown, which is effectively what a complaint or an audit looks for: “knowledge of the project, development/history of the project, input on earlier drafts, review of particular elements at earlier stages, or evidence of regular consultation throughout the project”. It expects the supervisor to be aware of the “relevant design criteria, methods of analysis, selection of materials and systems, field conditions, design constraints”, and says the supervised person “should not make independent engineering decisions without consultation and approval of the supervisor”.

NCEES frames the reason plainly: licensed professionals “must adhere to a strict code of conduct that requires them to practice in a manner that safeguards the health, safety, and welfare of the public”, and the statutes it quotes exist to “safeguard life, health, and property and protect the public welfare”. Which is why no delegation letter satisfies it. The seal is a personal attestation by a named human who can be disciplined and sued. Not transferable, and not divisible across a time zone.

A note on sourcing: the NCEES Model Law and Model Rules that many boards draw from were not openly retrievable when I checked. It matters little, because the model documents do not bind you. Your board’s statute does, and its definitions of responsible charge and engineer of record differ from the next state’s. Pull the rule text and read the section on sealing work prepared by others.

Review becomes engineering time, and that is where the saving goes

Here is the consequence that rarely appears in the business case. If the standard is being able to defend the design decisions, reviewing an offshore-produced calculation package is not checking. Checking verifies that numbers were transcribed correctly and the drawing matches the model. Defending means knowing why this beam depth and not one two sizes down, why that bearing pressure was accepted, why the assumed restraint at that support is real on site. To know that you have to reconstruct the decision, which means redoing enough of the work to hold an opinion about it.

On repetitive code-bounded work the reconstruction is quick, because the decisions are shallow and identical every time. On complex work it is slow, in proportion to the judgement that went into the design. That is the hidden cost, and it should be said without softening: on genuinely complex design, substantive review by the engineer of record can take a large enough share of the hours to erase most of the rate arbitrage. Not every project. But often enough that a business case modelling review at ten or fifteen per cent of production hours is built on a wish.

Two things follow. Review hours have to be budgeted and billed, which changes the fee proposal and not just the delivery model. And if the reviewing engineer keeps the same utilisation target and the same dates as before, they will eventually rubber-stamp. Not through laziness. Through arithmetic. A firm structured that way has decided to accept sealing risk, and the person carrying it is whoever’s name is on the sheet.

If a finished package lands on your desk with a request to seal it, the useful move is written and specific. Say you were not involved in the design development and cannot exercise responsible charge over it as it stands, then list what you need: the calculation basis, the assumptions log, access to the engineers who produced it, and review hours. Put a number on the hours. That turns a professional problem into a scheduling problem, which is a conversation your manager can actually have.

What genuinely travels, and what refuses to

The split is not about talent. It tracks how much of the work is executing a defined procedure versus forming a judgement about conditions the remote team cannot observe.

Travels well

Model production and drafting, given design intent and a standards package. High-volume, checkable, better with repetition.
Quantity take-offs and schedules. Rules-based, auditable against the model, and errors show up in the numbers rather than hiding.
Standard and typical detailing, where the firm has a real detail library rather than a folder of past jobs.
Routine calculation against a fixed code clause. Member sizing, deflection checks, drainage runs, load take-downs.
Clash detection and coordination. The strongest candidate: the tooling produces an objective issue list instead of an opinion, and overnight working becomes an advantage.

Does not travel

Code interpretation. Not application, interpretation: what to do when two clauses conflict, when the clause does not contemplate your condition, or when the authority having jurisdiction has a local reading that is not in the text.
Geotechnical judgement. Reading a log against local geology, deciding whether the investigation is adequate and what to assume where it is not.
Site-condition decisions. Existing structure, adjacent works, access, temporary works, anything where somebody had to have stood there.
Jurisdictional knowledge. Utility requirements, local amendments, permit review culture, the reviewer who always asks the same question.
Constructability trade-offs that turn on the local contractor market and on what this client will accept.

Code fluency, not language, is the real differentiator

The usual framing is that offshore delivery struggles on communication. That is the wrong diagnosis. English is generally fine, and written English in an issue register is often better than what the onshore team produces. The real gap is that designing to an unfamiliar jurisdiction’s code is a distinct professional competence, acquired over years.

It is not knowing a clause exists. It is knowing which clauses govern in practice, where the code is conservative and where it is not, which checks reviewers care about, and the body of convention that sits beside the code and is written down nowhere. An engineer fluent in one national code and handed another produces work that is traceable and quietly wrong in its priorities, a worse failure mode than obvious error because it survives a superficial check.

Firms that solve it do three specific things instead of assuming it away:

Training built on marked-up worked examples from the firm’s own completed projects, with the reasoning exposed, rather than a slide deck on clause numbers.
A code champion in the offshore team. One or two engineers who own the target code and keep a searchable log of every interpretation question and its answer. It concentrates the learning instead of spreading it thin, and it creates a career path that keeps the good people. Most firms answer the same twenty code questions for four years and capture none of them.
Reviewer pairing. The same engineer of record with the same offshore engineers across consecutive projects, so corrections teach rather than just fix. Rotating reviewers destroys this, which is why pooled staffing looks efficient and delivers badly.

A firm with no quality system onshore exports its own disorder

Distributed production does not create disorder. It reveals and multiplies whatever disorder was already tolerated. If your onshore team runs on tribal knowledge, undocumented standards and a senior engineer who fixes everything at the end, moving production offshore removes the only mechanism holding it together. What has to exist first:

A BIM execution plan that is actually followed: origins and shared coordinates, file naming, ownership per discipline, federation and issue cadence, software versions to the build number.
Written model and drawing standards: families, naming, annotation, line weights, view templates, what gets modelled versus annotated.
A defined level of development at each milestone, agreed element by element where it matters. A blanket “LOD 300 at DD” produces argument. A table saying what that means for foundations, slab penetrations and hangers produces work.
An issue register with named owners and dates, not an email chain. The highest-leverage artifact in distributed design, because it turns the time-zone gap into an overnight processing window.

If you would rather adopt a published framework than invent one, ISO 19650-2:2018 (information management using BIM, delivery phase of the assets) “specifies requirements for information management, in the form of a management process, within the context of the delivery phase of assets and the exchanges of information within it”. Both 2018 parts are under revision, so check status before naming an edition in a contract.

Liability and insurance: mechanisms to check, not rules to assume

General information follows, not legal or insurance advice. The point is which documents to pull and which questions to put in writing, since the answers depend on your policy, contract and jurisdiction.

Your professional indemnity policy. Find how it treats design performed by others, whether an affiliate, a foreign subsidiary or a subconsultant. Policies commonly distinguish work you perform from work you are vicariously liable for, and some address the territory where work is performed or where claims may be brought. Read the definitions, not the summary, then ask your broker in writing whether the arrangement is covered. Keep the reply.
Whether the offshore entity carries its own cover, and whether it can be reached. An intercompany transfer of work is a different risk position from a subconsultant with its own policy and an enforceable indemnity.
Your client contract. Clauses on key personnel, assignment and subcontracting of design, approval of subconsultants, where work may be performed, and data location. Territory restrictions do appear, and breaching one quietly is worse than declining the work.
Public-sector and publicly funded procurement conditions. Some attach requirements on place of performance, domestic preference, personnel security or data residency. They are programme-specific and they change, so read the solicitation rather than last year’s job.

Defence, critical infrastructure and some utilities work also carries data and export handling restrictions independent of engineering rules. One thing holds without checking any of it: none of these mechanisms move the seal. If your name is on the drawing, the board complaint arrives at your desk regardless of who was contractually liable to whom. Where there is no seal at all, as in the UK, the substance reappears as a named designer with defined duties, professional indemnity cover attaching to the designing entity, and the conduct obligations of the chartered engineer involved.

The junior pipeline, and why the frustration is not whining

The complaint in these threads gets dismissed as protectionism. A specific version of it is simply correct. Drafting and routine calculation are not low-value tasks that happen to fall to juniors. They are the mechanism by which judgement is acquired. You learn what a realistic member size looks like by sizing four hundred of them and being wrong about some. You learn how a building goes together by drawing the parts where it does not. Remove that rung and you do not get engineers who skipped the boring part. You get engineers who reach eight years of experience holding a review role with no basis for reviewing anything.

That risk lands on the firm years after the saving does. Anyone who can hold responsible charge over complex work is the output of a fifteen-year process, and a firm offshoring all production consumes the existing stock without replacing it. Mitigation costs money: work kept onshore deliberately for training rather than margin, juniors rotated through review alongside a senior rather than in place of one, and site time early, because nothing in production work substitutes for it. A firm that funds none of that has decided to buy senior engineers from other firms forever. That is a strategy, and worth naming as one.

What a working arrangement looks like

An integrated team, not a production line. One team across two locations, shared objectives, one issue register. The production-line version, where a client-facing team issues packages to a supplier team, guarantees that the people producing the work never learn why anything was decided.
Named individuals with continuity. The same engineers, project after project, known by name to the reviewing engineer. Ask for named seats in the agreement, with rotation notice and handover overlap. A pooled “we will resource it” means the learning curve restarts every job and you pay for it every job.
Protected overlap hours. Two to three hours of genuine daily overlap with the reviewing engineer available inside them, and questions answered within one working day. Shift one side’s pattern to get it, and treat the overlap as a commitment rather than a courtesy.
The engineer of record involved before production starts. Design criteria, assumptions, calculation approach and key details set with the reviewer in the room at the outset, then review at early stages rather than only at the end. Involvement, in the guideline’s words, is shown through “input on earlier drafts, review of particular elements at earlier stages, or evidence of regular consultation throughout the project”. Front-loading is not good-practice advice here. It is what makes the seal defensible.
A measured rework rate. Comments per sheet, or rework hours as a percentage of production hours, tracked per package and per team over time. It is the only number that shows whether capability is being built or whether you are running a permanent correction service. If it is not falling across consecutive projects, the training and pairing are not happening.

None of that is exotic, and it is absent from most arrangements. Which is why the complaint in these threads is usually about a real defect rather than about geography.

If your conclusion is that you need engineers embedded in your own team and accountable to your own engineer of record, rather than a production shop issuing packages over a wall, that is the model AB7 Solutions staffs: contract staffing and staff augmentation for design, BIM and drafting roles, filled as named remote professionals who join your standups, your federated model, your issue register and your review cadence, with rotation notice and handover overlap agreed in writing, because continuity is what makes the reviewing engineer’s job possible. We also build the surrounding machinery when it is missing: standards and documentation work, and workflow automation for issue tracking and drawing transmittals so the register stops living in email. Send us the package split you are considering and we will tell you which parts suit a remote seat and which should stay with your licensed staff. If the honest answer is one senior local hire instead of a team from us, we will say so. Call +1 321 341 7733, email ab@ab7solutions.com or director@ab7solutions.com, or start at www.ab7solutions.com.

Sources: NSPE, Code of Ethics for Engineers (II.2.a, II.2.b, II.2.c on competence, signatures, and documents prepared under an engineer’s direction and control); Engineers Canada, Public guideline on direct supervision (what supervision requires, how active involvement is demonstrated, and the limits on independent decisions by supervised staff); NCEES, Why licensure matters (code of conduct and the public-protection purpose of licensure statutes; the NCEES Model Law and Model Rules were not publicly retrievable at the time of writing, and each board’s own statute governs regardless); ISO, ISO 19650-1:2018 and ISO 19650-2:2018 (concepts and principles, and delivery-phase requirements; 2018 editions under revision). Liability, insurance and procurement points are general information, not legal or insurance advice.