Drones, the unmanned aerial vehicles (UAVs) that fly inspection, delivery, survey and agriculture missions, have crossed from hobby hardware into a regulated commercial fleet. The FAA counted roughly 966,000 registered commercial small drones by the end of 2024 and forecasts the fleet past one million in 2025, toward 1.18 million by 2029 . The proposed Part 108 rule would normalize beyond-visual-line-of-sight operations at or below 400 feet, moving the industry from waivers to routine approvals for delivery, agriculture and aerial surveying . Hiring for this discipline means staffing the flight control, drone navigation, drone perception and certification work behind that fleet.
Challenges in Drones Recruiting
Unmanned aerial vehicles (UAVs) outgrew the waiver model
For a decade, commercial drone work lived inside Part 107 and its visual-line-of-sight rule, and everything beyond it ran through individualized waivers and exemptions. The FAA's proposed Part 108 replaces that model with a performance-based framework: operators apply for a permit or a certificate, aircraft up to 1,320 pounds gain airworthiness acceptance against industry consensus standards instead of traditional certificates, and routine BVLOS operations stop being exceptions . Operations of multiple aircraft, the backbone of swarm drone systems and delivery fleets, would still be evaluated case by case .
The regulatory shift changes what employers are actually hiring. Under the waiver model, a strong pilot with paperwork skills could carry an operation. Under Part 108, operators must stand up risk assessments, manuals, training programs and, for certificated paths, a safety management system . Demand moves toward engineers who can produce safety evidence, not just fly. Programs that staffed for the waiver era and are now hiring for the rulebook era discover the two profiles are different people.
Drone autonomy grades into certification risk categories
Europe sorts the same technology question through EASA's categories. Operations outside the open category's limits enter the specific category, where the SORA methodology walks a ten-step process from ground risk class and air risk class to a SAIL level and 24 operational safety objectives . The higher the SAIL, the heavier the design evidence: SAIL IV needs an EASA design verification report, and SAIL V and VI need a type certificate under Part 21 . The SORA 2.5 package adopted by EASA in September 2025 updates that methodology, adding quantitative ground risk modeling while the agency explicitly kept cybersecurity out as disproportionate .
Drone autonomy hiring inherits this grading. An autonomy engineer whose work stops at a declaration against design objectives is not interchangeable with one who has pushed a design through a design verification report. The algorithms look similar on paper; the difference is robustness evidence, containment analysis and a relationship with an authority that survived review. Recruiters who screen for SLAM and planning keywords get the first profile when the operation needed the second.
Flight control splits hobby stabilization from certifiable aircraft
Flight control spans a wide gap. At one end sits the open-source autopilot world: attitude loops, control allocation, failsafe return-to-home, waypoint following on PX4-class stacks. At the other end sits the Part 108 airframe: an aircraft of up to 1,320 pounds carrying Remote ID and lighting, held to detect-and-avoid obligations that require it to yield to ADS-B Out traffic, with strategic deconfliction and conformance monitoring in controlled airspace . The FAA's concept of operations expects DAA standards to absorb even degraded ADS-B performance .
The hiring risk is that both populations write flight control on their CVs. One has tuned a quadcopter in a field; the other has validated envelope limits, link-loss behavior and right-of-way logic against a safety case. The question that separates them is small: which failure modes the candidate designed for, and which evidence they produced that the design worked. Programs building toward Part 108 or EASA certified-category work need the second profile, and it is the rarer one.
Drone navigation divides GPS-dependent and GNSS-denied pipelines
Drone navigation splits on the availability of a satellite fix. Precision agriculture and surveying lean on RTK-GNSS for centimeter-level positioning. Indoor and bridge inspections, urban canyons and below-canopy flight do not have that luxury, and the field has moved toward visual-inertial and lidar-inertial odometry instead. Research at the SPRIN-D Funke challenge pushed a fully onboard system through kilometer-scale waypoint navigation below 25 meters without GNSS, correcting odometry drift by matching lidar heightmaps against prior geodata on CPU-only hardware . Visual-inertial odometry in GPS-denied indoor flight still fights scale error and long-term drift, the problems landmark-based correction methods exist to fix .
These are two different engineers. The GNSS side is about base stations, RTK corrections, integrity and georeferencing; the denied side is about drift, loop closure and sensor fusion under vibration. A brief that says drone navigation without saying which environment will collect both, and only one of them will fit the mission.
Drone perception runs against weight and power budgets
Drone perception is constrained in ways ground robotics is not. Every sensor gram subtracts endurance, and the mission decides the stack: cooperative traffic detection reads ADS-B, while non-cooperative detect-and-avoid needs cameras, radar or lidar running onboard, sometimes against uncooperative lighting and featureless sky . Below a certain airframe size the compute budget forces sensor fusion onto edge hardware, which is why aerial robotics engineers spend so much effort on what runs where.
The candidate pool for this work is thin because the experience is specific. Automotive perception engineers know lidar and cameras but not airframe vibration, mounting isolation and the altitude where the horizon stops helping feature tracking. An aerial robotics team that survives its first real deployments has usually learned, through flight test, lessons that no adjacent industry teaches. Hiring from that pool means screening for airframe hours, not just model names.
Aerial robotics claims need flight logs, not demo footage
Assessment in drones fails when the interview stops at the demo reel. The discipline is documented by regulation: every real operation carries an authorization or waiver class, an operating volume, flight hours, weather and link-loss envelopes, and a record of what went wrong. The probes write themselves. Ask which authorization class the candidate flew under, what the SORA said about ground and air risk, what happened when the link dropped at the far edge of the volume, and what the battery-endurance margin was on the longest real flight . Ask to see the flight logs.
Candidates who answer in numbers and authorizations have run operations; candidates who answer in footage have watched them. The cost of guessing wrong lands in approvals: a denied authorization, a lost waiver, or an incident that grounds the fleet while a regulator re-reads the safety case. In a discipline moving from waivers to a rulebook, the engineers who can produce evidence are the ones who keep the fleet airborne, and they are a small population worth screening hard for.
References
- FAA Aerospace Forecast Fiscal Years 2025-2045 — Federal Aviation Administration (FAA). (accessed 2026-09-28)
- Normalizing Unmanned Aircraft Systems Beyond Visual Line of Sight Operations (Proposed 14 CFR Part 108) — Federal Aviation Administration (FAA), Federal Register. (accessed 2026-09-28)
- Specific Operations Risk Assessment (SORA) — European Union Aviation Safety Agency (EASA). (accessed 2026-09-28)
- ED Decision 2025/018/R: Regular Update of the AMC and GM to Commission Implementing Regulation (EU) 2019/947 — European Union Aviation Safety Agency (EASA). (accessed 2026-09-28)
- Kilometer-Scale GNSS-Denied UAV Navigation via Heightmap Matching — arXiv (2510.01348). (accessed 2026-09-28)
- Drone Integration: Concept of Operations (May 2025) — Federal Aviation Administration (FAA). (accessed 2026-09-28)
- Landmark-Based Scale Estimation and Correction of Visual Inertial Odometry for VTOL UAVs in GPS-Denied Environments — MDPI Sensors. (accessed 2026-09-28)