A first engineering team in Sofia, without an entity

Template — replace with a real engagement. The structure below is what converts; substitute your own client, numbers and timeline before publishing.

The client builds control software for industrial equipment. Around 400 people, all of them in Germany, and a product roadmap that had been slipping for two years because the local market could not supply embedded and backend engineers at a price the business would approve. The board had already agreed on Bulgaria in principle. What they did not have was a legal entity, a payroll provider, an office, or anybody internally who had hired outside Germany before.

The binding constraint was time. They wanted engineers writing production code within a quarter. Registering a company, getting on the revenue authority’s books, appointing an accountant and signing an office lease is a multi-month path on its own, and it has to finish before the first employment contract can even be drafted. They also wanted an exit. If Sofia did not work, they did not want to spend a year unwinding a subsidiary to prove it.

There was a softer problem underneath. Their engineering managers were sceptical. They had lived through an offshore outsourcing project years earlier that produced code nobody wanted to maintain, and they assumed Bulgaria would be the same thing with a shorter flight. Nothing about the commercial model would fix that; only working with the actual people would.

We recommended the outsourced-team model over permanent placement, and we said so before fees came up. With no entity here, direct hiring was closed to them for at least two quarters. On our payroll, the first engineer could start on a normal Bulgarian employment contract while the client kept complete control of the work, the tooling and the code.

How the search was run

• Calibration before sourcing. Two working sessions with their lead architect to turn German job descriptions into something a Sofia engineer would recognise — including dropping a formal degree requirement that would have cut the candidate pool roughly in half for no gain.
• A market read, in writing. Salary bands per profile at the seniority they actually needed rather than the seniority they had written down, plus a plain note on which of the roles was going to be the hard one and why.
• Direct approach, not adverts. The engineers they needed were employed and not applying anywhere. Sourcing was outbound, with the client named only once a candidate had signalled real interest.
• Their managers in the room from week one. The client’s own engineering leads ran the technical interview themselves. That did more for the scepticism than any reference we could have supplied.

We also carried the parts that are not recruitment: employment contracts, payroll, social and health contributions, equipment, and desks in a Sofia co-working space until the team was big enough to justify its own floor. The client received one monthly invoice and one named account manager, not a vendor portal.

The first engineers started TBC weeks after the calibration session, and the team reached TBC people over the following TBC months. The role we had flagged as difficult took TBC weeks longer than the others, which is close to what we told them to expect in the first conversation.

Two things mattered more than the headcount. The first is that the German managers stopped calling Sofia “the external team” somewhere around month TBC; by then sprint planning was joint and code review ran in both directions. The second is that when the client eventually did incorporate in Bulgaria, TBC of the team transferred onto their own payroll. That is the intended end state of this model, not a loss for us. Leasing is a way to start, not a way to stay.

Before publishing: replace this paragraph with what actually went wrong on the engagement and how you handled it. A case study in which nothing went wrong reads as marketing and buyers discount it.