Four competence families behind a drone built at rate, read from artefacts

The method, stated before the content

This dossier derives competences from shipped artefacts, published certifications and dated announcements. It does not use job-board listings, recruitment aggregators, or estimated headcounts. That restriction is deliberate: aggregator pages are unstable, frequently unattributed, and rewritten without notice, and Delair's own headcount is reported in the open record with three incompatible values (84, 217 and 399), which means no staffing-derived inference about its skills mix can be trusted. What cannot be broken, however, is the logic that a company which has flown a beyond-visual-line-of-sight mission employs people capable of getting one authorised. That is the inference chain used throughout.

Family I — Attritable production engineering

The Boussens expansion, taken over from an automotive-grade manufacturer and framed publicly around a rate of about 100 units a day, is the single most demanding competence statement the company has made. It is not an aerospace-craft skill set. Producing an attritable aircraft at that cadence requires industrialisation engineers who can design fixturing and assembly sequences for a configuration that will change; test engineers who can automate functional and radio-frequency acceptance rather than perform it by hand; supply-chain engineers who can qualify a second and third source for parts whose first source may be export-controlled or simply sold out; and manufacturing-quality staff who can hold traceability at speed.

Note the specific tension a candidate should understand before applying. In classical aerospace, a configuration change is a slow, documented event. On an attritable line responding to electronic-warfare pressure, configuration change is the steady state. The scarce competence is therefore not "serial production" in general — it is serial production under continuous engineering change, with a quality system that survives it. The published ISO 9001 certification is the formal expression of that: someone has to own the documented process, the internal audits and the non-conformance loop while the product moves underneath them.

Family II — Certified and authorised flight operations

The DT26's beyond-visual-line-of-sight record and the existence of a C6 class-marked configuration are regulatory artefacts, not engineering ones. Reaching them requires competence in the European operating framework: operational risk assessment, ground-risk and air-risk analysis, containment argumentation, operations manuals, remote-pilot competence records, and a working relationship with the national aviation authority.

This is the least visible and most transferable skill in the sector. The engineer who can demonstrate containment for a corridor inspection flight over inhabited terrain holds a competence that generalises across operators, industries and countries, because the framework is European rather than company-specific. We flag one documentary caution: the release date of the C6 configuration is stated inconsistently across the sources that mention it, and we print that disagreement rather than resolving it. The competence conclusion does not depend on the date.

Family III — Autonomy, perception and confined-space navigation

The Squadrone System acquisition is the cleanest evidence in this dossier for a specific software competence, because Squadrone's product line was rotary-wing indoor inventory — flight in exactly the environment where satellite navigation is unavailable. That means visual-inertial odometry, onboard perception, obstacle avoidance, and mission autonomy that continues to function without an external position fix. Roughly 22 people carried that competence into the group at the time of the acquisition.

Placed beside the fixed-wing and tactical platforms, the industrial logic becomes explicit: navigation-denied autonomy is no longer an indoor-logistics niche. It is the answer to jamming. A candidate reading this should understand that competence in GNSS-denied navigation, terrain-relative or visual guidance and onboard inference is not a research speciality in this market — it is a production requirement, and its natural European talent pools are exactly the robotics and computer-vision laboratories of the Grenoble and Toulouse basins.

Family IV — Programme, export and defence-grade governance

The step from DT46 to Colibri and Damoclès is a governance step before it is a technical one. A company that publishes loitering munitions in its range, delivers to a country at war and participates in France 2030 and the VOLTAIR programme necessarily employs people who handle export-control classification and licensing, defence-customer programme management, and the arming-chain and handling disciplines that a survey-drone business never needed.

This family is systematically under-described in public discussion of the drone sector, and it is where mid-career entrants from established primes have the clearest advantage. The competences are procedural and legally consequential: an incorrect export classification is not a delay, it is an offence. Equally, dual-use industrial-policy funding brings audit and reporting obligations that require staff able to write to a public funder's standard.

What a candidate should verify before applying

We would be failing the reader if we presented this as a settled picture. Three things about Delair are genuinely uncertain in the open record, and anyone considering the company should ask about them directly rather than trusting a published figure.

• Scale. Three headcounts circulate in the openable record — 84, 217, 399. They almost certainly describe different perimeters. Ask which entity, which date, and whether the figure is full-time equivalents.
• Revenue trajectory. One openable 2025 figure exists, about €50¹m, reported by regional press in July 2026. Ask whether it is audited and on which perimeter.
• Ownership. Claims attaching Delair to a "Groupe Chiron" led by Michael Kluehr, to Tikehau Capital, or to Ace Capital Partners are unsupported by any document we could open. We state no shareholder. A candidate weighing job security should ask who controls the company and expect a documented answer.

How to read the rate objective as a candidate

The single most instructive line in the Delair record is not a technology claim: it is the choice, announced on 19 June 2026, to industrialise inside an automotive supplier's plant rather than to build a new one. Read as a labour-market signal, that decision says the missing competence was never aerodynamics — it was serial manufacturing discipline. Automotive assembly brings takt time, statistical process control, supplier qualification and a change-order regime that a defence prototype shop rarely runs. A candidate who has lived that regime in a car or bearing plant is, on this evidence, closer to the scarce profile than an aerospace generalist.

The second instruction is to treat the stated hundred-per-day figure as an objective with a date attached, not an output. It is the yardstick a new hire will be measured against, and it defines the questions worth asking in an interview: which parts are single-sourced, what the first-pass yield is, who signs a design change once a unit is fielded, and how a qualified configuration is held stable while the threat moves. The ISO 9001:2015 certificate proves a documented system exists; it says nothing about the rate that system has actually sustained. Ask for the rate.

What we refuse to print

• No salary bands, no compensation ranges, no benefits. Nothing verifiable exists in the record we consulted.
• No hiring volumes, open-role counts or per-site staffing. Not published.
• No named tools, languages or frameworks beyond what shipped products and certifications imply. We will not reconstruct a technology stack from a recruitment page we cannot authenticate.
• No claim about internal mobility, training programmes or attrition. Absent from the record.
• No Blue UAS status and no Donecle acquisition: neither is supported by a document we opened, so neither shapes the competence map.

The strategic conclusion

The competence profile that emerges is unusual, and it is worth naming precisely because career advice in this sector tends to default to "learn to fly drones". That is not the constraint. Europe has no shortage of people who can design and operate a small aircraft. What it is short of is people who can hold a quality system together while the product changes weekly, argue an operational-risk case to an aviation authority, make an aircraft navigate without satellites under jamming, and classify a controlled export correctly the first time.

Those four capabilities have almost nothing in common with each other — which is precisely the point. The scarce professional in this market is not a specialist in one of them but a person who can work across two, because the interfaces between them are where European drone programmes actually stall: between the production line and the change order, between autonomy research and airworthiness argument, between a war-time delivery and an export licence. Anyone building a career here should choose their second competence deliberately, and should ask an employer to show, in documents, the scale and the ownership it is asking them to bet on.

Reporting standard: every competence above is derived from a shipped product, a published certification or a dated announcement we opened. No figure is estimated. Contradictions are printed as contradictions. Absences are named as absences, not filled.