Should I Hire an MBSE Expert or Train My Existing Engineers?

For most teams, train. Your engineers already own the hardest, slowest thing to learn — your domain — and SysML is the learnable part. Hiring makes sense for a narrow set of needs: method decisions, an adoption kickstart, and periodic expert review. Those needs are better met with advisory hours than with a full-time unicorn req that sits open for two quarters.

That’s the answer. Here’s the math and the exceptions, because this decision is usually made on instinct and it’s expensive to get wrong in either direction.

The req that won’t fill

Start with an observable fact: open MBSE reqs sit. A posting that asks for SysML depth, Cameo fluency, your specific domain, and (in defense) a clearance is describing a few hundred people in the country, most of whom are happily employed and being recruited by primes with deeper pockets than yours. Three to six months of vacancy is normal; nine isn’t rare. Every month that req is open, the modeling work it was supposed to do either doesn’t happen or lands on someone already at capacity.

So the honest comparison isn’t “hire vs train.” It’s “train now vs maybe-hire in Q3, then ramp them for six months after that.”

What your engineers already have

The three things an effective modeler needs are tool fluency, your model’s conventions, and domain knowledge — we broke these down in Why Does a New MBSE Hire Take Six Months to Become Productive?. Your existing engineers hold the third outright and half of the second. They know why the design is shaped the way it is, which interfaces are contentious, which requirements have bodies buried under them. That knowledge takes years to build and cannot be trained in.

What they’re missing is SysML and Cameo. That is a teachable skill with a known curriculum. An engineer who can already think in systems learns enough SysML in a focused week to start doing real, supervised work, and reaches independence in a couple of months of practice. We’ve never seen it go the other direction that fast — a tool expert learning avionics, or propulsion, or your program’s fifteen years of design rationale, in two months.

The math sketch

Round numbers; substitute your own.

Hiring an MBSE expert:

• Recruiter fee: 20–25% of a ~$150K salary — call it $30K–$38K
• Vacancy: 3–6 months of the work not getting done
• Ramp after start: ~6 months to full productivity, roughly $60K–$70K of loaded cost at partial output
• Result: one productive expert, roughly 9–12 months from opening the req, and a new single point of failure

Training a cohort of five:

• Instructor-led SysML fundamentals for five engineers: low tens of thousands, all-in
• Advisory support while they apply it (a few hours a week for a quarter): comparable again
• Productivity dip during training: a week of class plus part-time practice — real, but small and scheduled
• Result: five people at working proficiency inside a quarter, redundancy instead of key-person risk, and the domain knowledge never left the building

The cohort path typically lands at a third to a half of the hiring path’s first-year cost, arrives two or three quarters sooner, and — this is the part the spreadsheet undersells — produces a team instead of a hero. When one trained engineer leaves, you still have four.

When you genuinely need outside expertise

Training isn’t the whole answer, and pretending it is would be its own mistake. There are three jobs your newly trained cohort can’t do for themselves:

• Method decisions. Package structure, diagram usage policy, how requirements trace, what “done” means for a model. Get these wrong at the start and every hour of modeling afterward compounds the error. This is judgment built from seeing many programs, not from one course.
• The kickstart. The first month of a new modeling effort sets patterns everyone else copies. An experienced modeler establishing the skeleton — the conventions, the seed model, the first worked examples — makes the whole cohort faster.
• Review cadence. A monthly expert pass over the model catches drift while it’s cheap to fix. Your team gets steadily better because every review is also a lesson.

Notice the shape of all three: they’re measured in hours per week, not heads. Two to four advisory hours a week buys you the judgment of someone who’s done this repeatedly, without the recruiter fee, the vacancy, the ramp, or the risk that your unicorn gets poached in year two. Buying a full-time expert to cover part-time expert needs is how programs end up with an expensive modeler doing commodity diagram work.

The hybrid that actually works

The pattern we recommend to most teams under about fifty engineers:

• Train a cohort — three to six of your own people, together, on your calendar
• Bring in advisory hours for method decisions, the kickstart, and monthly model reviews
• Write the conventions down as you go, so the next cohort ramps even faster
• Revisit the hire question in a year — if modeling demand has grown past what the cohort covers, you’ll now write a better req and interview with far sharper questions

If you do eventually hire, hire into a team that already models, with conventions already written. That hire ramps in weeks, not months, because two of the three ramps are paved.

Where to go from here

If you’re weighing this decision for your team, CameoMagic does both halves of the hybrid: instructor-led SysML training for cohorts, and advisory engagements sized in hours per week rather than heads. The no-risk way to start is a free 30-minute model review — show us where your program is, and we’ll tell you honestly whether your situation calls for training, advisory hours, or a hire. Book it at cameomagic.com.