A corpus, built and maintained in public, as proof of the architecture.

Unmanned aerial vehicles operating across the size spectrum from sub-kilogram quadcopters to Group 5 fixed-wing platforms with wingspans exceeding 50 feet. Includes both fully autonomous platforms and operator-in-the-loop systems where autonomy reduces operator load. Excludes commercial passenger aviation (covered under eVTOL) and consumer drones below the prosumer line.

The segment splits operationally into three pools that draw on overlapping but non-identical talent: defense-mission UAS (ISR, strike, attritable platforms), commercial UAS (inspection, logistics, mapping), and counter-mission UAS (which sits in its own segment, see 05).

VP Engineering, Director of Engineering, Head of Flight Software, Head of GNC (guidance, navigation, control), Head of Avionics, Principal Flight Software Engineer, Principal Embedded Systems Engineer, Senior Staff Flight Test Engineer, Founding Engineer (early Series A), Director of Manufacturing & Production Engineering, Head of Mission Systems.

Companies in this segment hire flight software ahead of avionics, and avionics ahead of manufacturing. The first 25 engineering hires skew heavily toward flight software, GNC, and embedded systems. The shift to manufacturing and production engineering happens at the Series B inflection, typically 12-18 months after Series A close. Defense-mission UAS companies add mission systems engineers earlier than commercial UAS companies because defense customers will not accept a platform without one.

The segment compounds: Anduril alumni have founded the next generation of US drone companies, who are now hiring the next generation of operators, several of whom will themselves found drone companies in 24 months. The capital flow is reinforcing. Companies that don't hire ahead of the alumni-network curve get out-recruited by ones that do.

Uncrewed surface vessels (USVs) and uncrewed underwater vehicles (UUVs) operating in defense, scientific, and commercial maritime contexts. Includes both small-form-factor platforms (man-portable, ship-launched) and larger purpose-built autonomous vessels. Excludes manned vessels with assistive automation.

The defense subsection has matured fast since 2023. Saronic's $4B valuation in 2024 marked a step change. Commercial maritime autonomy lags but is starting to follow.

VP Engineering, Director of Naval Architecture, Head of Marine Systems, Head of Autonomy (maritime-specific), Principal Subsea Engineer, Senior Staff Hydrodynamics Engineer, Director of Sea Trials & Operations, Head of Sensor Integration, Mission Systems Engineer (maritime-specific), Principal Communications Engineer (RF in maritime contexts).

Maritime autonomy hiring is bottlenecked by naval architecture talent: there are far fewer trained naval architects than software engineers, and the discipline can't be cross-trained into from adjacent fields on a 12-month timescale. Companies that lock down naval architecture leadership early outperform companies that try to hire it on demand. Subsea hiring follows surface hiring by 6-12 months as platforms mature.

This segment is the smallest of the Tier 1 vehicle classes by company count but among the highest by capital concentration. Defense maritime is structurally preferred by US DoD over commercial maritime for procurement reasons. Companies entering the segment from a defense-first frame outperform companies entering from a commercial-first frame on capital efficiency.

Uncrewed ground vehicles (UGVs) and autonomous-driving systems in both defense and dual-use commercial contexts. The segment includes light tactical UGVs, autonomous logistics vehicles (military and commercial), autonomous trucking, and platform-agnostic autonomous-driving software stacks. Excludes consumer-vehicle ADAS (out of scope) and warehouse robotics (covered under industrial robotics, B-tier adjacent).

VP Engineering, Head of Autonomy Stack, Director of Vehicle Engineering, Head of Perception (vehicle-specific), Principal Motion Planning Engineer, Senior Staff Behavior Engineer, Director of Safety Engineering, Head of Hardware-in-the-Loop Test, Principal Sensor Integration Engineer, Head of Mapping & Localization.

The autonomous-trucking subset commands a 10-15% premium on these bands due to talent compression.

Defense ground autonomy hires safety engineering later than commercial ground autonomy. DoD acceptance of "autonomy with operator override" is more permissive than civilian regulator acceptance of full self-driving. Commercial autonomous trucking, by contrast, hires safety engineering and verification almost as early as core autonomy engineering. Companies that mismatch the hiring sequence to their actual customer (defense vs. civilian) burn capital on the wrong sequence.

The autonomous-driving segment went through a capital correction in 2023-2024 that pushed talent out of struggling companies and into adjacent segments, particularly defense ground autonomy and autonomous trucking. The talent quality available right now in 2026 is unusually strong relative to the funding available, which creates an arbitrage for well-capitalized Series A and B companies hiring at this moment.

Electric vertical takeoff and landing aircraft for passenger and cargo applications, including urban air mobility (UAM), regional air mobility (RAM), and cargo-eVTOL platforms. The segment includes both clean-sheet eVTOL designs and electrified conversions of conventional rotorcraft. Excludes pure UAS (covered in 01) and traditional rotorcraft.

The segment is at the inflection between certification and commercial operation. The companies that close this transition successfully will define the segment for the next decade.

VP Engineering, Head of Certification, Director of Flight Test, Head of Powertrain Engineering, Principal Power Electronics Engineer, Senior Staff Aerodynamicist, Head of Avionics, Principal GNC Engineer, Director of Aircraft Systems, Head of Manufacturing Engineering.

Certification is the bottleneck. The number of engineers in the world with eVTOL-relevant FAA Part 23 / EASA SC-VTOL certification experience is small enough to be effectively a known list. Companies that hire certification leadership early raise faster on subsequent rounds because investors price in lower regulatory risk. Companies that try to backfill certification at Series B regularly slip 12+ months on certification timelines.

Power electronics talent is the second bottleneck. The discipline cuts across automotive EV, eVTOL, and electrified marine. Companies that don't compete on equity and mission against the EV sector lose the talent fight.

The eVTOL segment is unusually concentrated by capital. Five companies (Joby, Archer, Volocopter, Lilium successor entities, Vertical Aerospace) account for the majority of disclosed equity capital raised in the segment. Series A and B companies in eVTOL are competing for talent against companies that have raised 10-50× more. This means the smaller companies need to compete on something other than comp: typically mission specificity, technical interest, or equity weighting. Recruitment in this segment that ignores this dynamic underperforms.

Detection, tracking, identification, and neutralization systems for hostile uncrewed aerial threats. The segment includes RF/radar/EO sensor systems, mission software for threat assessment, and effector platforms (kinetic, electronic warfare, directed energy). Excludes general air defense (covered under traditional defense primes).

The segment has compressed dramatically since 2023 as drone use in active conflict has shifted procurement priorities. Capital and procurement attention have both tilted hard toward US-based counter-UAS startups in the last 18 months.

VP Engineering, Head of Sensor Engineering, Head of Mission Software, Principal Radar Engineer, Principal RF Engineer, Senior Staff Computer Vision Engineer (drone-detection-specific), Director of Threat Assessment Algorithms, Head of Effector Integration, Principal Electronic Warfare Engineer.

Radar engineering is the structural bottleneck. The talent pool was built around a small number of legacy defense primes and has not been replenished at the rate required by the current wave of counter-UAS startups. Companies that hire radar leadership early can attract the rest of the team around that anchor; companies that don't, can't.

Mission software hiring follows sensor engineering by 6-9 months. Companies that try to build the mission software layer ahead of the sensor layer overspecify the software against assumptions about sensor capabilities that turn out to be wrong.

The segment is the most clearance-sensitive of the Tier 2 categories. Roughly 60-70% of senior engineering roles require active or eligible US security clearance. Clearance status functions as a hard partition: the cleared and non-cleared pools price and move differently, and analysis that ignores the split misreads the market.

The software layer that turns autonomous platforms into operational capabilities: mission planning, command-and-control (C2), data fusion across multiple platforms, operator interfaces, and the AI/ML systems that surface decisions to operators. Includes both startup-built mission software (Anduril Lattice, Shield AI Hivemind) and the broader category of platform-agnostic mission tools.

Mission software is where defense autonomy companies are increasingly differentiated. The platform layer has become commoditized faster than the mission software layer, and capital is following the differentiation.

VP Engineering, Head of Mission Software, Director of Platform Engineering, Principal Distributed Systems Engineer, Senior Staff Backend Engineer (with defense data domain experience), Head of UI/UX (defense-context-specific), Principal AI/ML Engineer (decision systems), Head of Data Engineering, Director of Integration Engineering.

Mission software companies hire backend and distributed systems engineers earlier than platform-engineering or UI/UX. The first 30 engineering hires are typically 70%+ backend / distributed systems. UI/UX hiring tends to lag, and the lag is a recurring failure mode. Defense customers are notoriously sensitive to operator-interface quality, and companies that under-invest in UI/UX get returned by user testing late in development cycles.

The Anduril effect compounds in this segment more than in any other Tier 2 category. The Lattice platform has redefined what defense customers expect from mission software, which raises the floor for every other entrant. Companies entering the segment now have to clear a quality bar that was lower 24 months ago. This compresses the talent market: the same engineers who could have been first-50 hires at Anduril in 2022 are being recruited to start companies in the same segment in 2026.

The full sensor pipeline from raw sensor input through perception output: lidar, radar, camera, GNSS-denied positioning, multi-sensor fusion, scene understanding, object detection and tracking. The segment cuts across vehicle classes: perception engineers serve aerial, maritime, ground, and space autonomy contexts, and the underlying skills transfer cleanly between them, though deployment specifics differ.

VP Engineering (perception-led), Head of Perception, Director of Sensor Engineering, Principal Computer Vision Engineer, Principal Sensor Fusion Engineer, Senior Staff Lidar Engineer, Principal Radar Algorithms Engineer (perception-side), Head of Calibration & Sensor Models, Principal SLAM Engineer, Head of Scene Understanding.

Perception is the most cross-segment-fluid talent pool in autonomous systems. A senior perception engineer working on autonomous trucking can move to maritime autonomy or eVTOL with 3-6 months of ramp. This makes the segment unusually competitive: perception leadership at any single company is constantly being recruited by every other autonomy company hiring perception leadership. Retention is structurally hard.

The corollary: perception bench depth is the single strongest recruitment moat in autonomous systems. Companies that maintain pre-qualified perception benches across vehicle classes can fill perception roles in weeks where competitors take quarters.

This segment has the strongest cross-vehicle-class transfer of any Tier 2 or Tier 3 category. The implication for anyone mapping this market: a map organized by vehicle class loses information a map organized by discipline keeps. This corpus organizes perception by discipline first, then by application context, a structural choice the transfer pattern forces.

The decision-making layer of autonomous systems: guidance, navigation, and control (GNC) for aerospace platforms; motion planning and behavior generation for ground and maritime platforms; high-level mission planning and autonomy arbitration. The segment is the connective tissue between perception (what the system sees) and actuation (what the system does).

Head of Autonomy Stack, Principal GNC Engineer, Principal Motion Planning Engineer, Senior Staff Behavior Engineer, Principal Control Theory Engineer, Senior Staff Reinforcement Learning Engineer (autonomy-applied), Director of Autonomy Architecture, Head of Decision Systems.

Autonomy stack hiring is bottlenecked by control-theory talent at the senior and principal levels. Strong control engineers are not interchangeable with strong ML engineers, despite a category of confusion that emerged in 2023-2024. Companies that hire ML engineers and ask them to do control engineering ship buggy systems; companies that hire control engineers ahead of the curve outperform.

The architecture of the autonomy stack tends to ossify around the first principal hires. Companies that hire well in the first 5-7 autonomy stack roles benefit for years; companies that hire poorly carry the architectural debt for years. This is one of the highest-leverage hiring sequences in autonomous systems.

The senior end of this segment is one of the smallest globally: true principal-level GNC and control-theory engineers number in low thousands worldwide, not tens of thousands. Companies that maintain warm relationships with this population over multi-year horizons (rather than only contacting them when a role opens) get the early signal when senior people consider moves.

Machine learning systems that run on resource-constrained hardware in autonomous platforms: onboard inference for perception, decision, and control loops, model compression and quantization, custom inference accelerators, and the integration of ML into real-time embedded systems. Excludes pure cloud-side ML (out of segment) and consumer edge ML (out of segment).

Head of Edge ML, Principal Embedded ML Engineer, Senior Staff Model Optimization Engineer, Principal Hardware-Software Integration Engineer (ML focus), Director of Inference Infrastructure, Principal ML Compiler Engineer, Senior Staff Deep Learning Engineer (embedded systems experience).

Edge ML hiring is constrained by the intersection of two relatively scarce skill sets: strong ML engineers who also have deep embedded systems experience. The Venn diagram is small. Companies that try to hire each side separately and integrate them through process tend to ship slower than companies that hire engineers who span both. The latter cost more, hire harder, and ship faster.

This segment is where the talent scarcity is most acute relative to demand right now. Capital flow into autonomous platforms outpaces edge ML talent supply by a wide margin. The result: edge ML hiring is taking 6-9 months for senior roles even at well-funded Series A and B companies.

The hardware accelerator generational shift (Jetson Thor, custom silicon, NPU integration) has created a 2026-specific dynamic: engineers with experience on the previous generation of hardware are hireable but require 3-6 month ramp to be productive on current platforms. Companies that screen on prior platform experience without considering ramp are filtering out a meaningful percentage of the actually-good talent pool. A screen built on ramp trajectory rather than historical platform mix keeps that pool in view.

The full lifecycle of autonomous-system verification and validation: high-fidelity simulation environments, hardware-in-the-loop test, scenario coverage analysis, safety case construction, formal verification, and certification engineering for both aviation (FAA, EASA) and ground-vehicle (NHTSA, UNECE) regulators. The segment cuts across vehicle classes but is most weight-bearing in eVTOL and autonomous trucking, where regulatory acceptance is the bottleneck to commercial deployment.

Head of Verification & Validation, Director of Safety Engineering, Principal Simulation Engineer, Senior Staff Hardware-in-the-Loop Test Engineer, Head of Certification, Principal Safety Case Engineer, Director of Test Operations, Senior Staff Formal Verification Engineer, Principal Coverage Analysis Engineer.

This is the most under-staffed engineering function in autonomous systems at Series A and B. Founders prioritize hiring the building functions (perception, autonomy stack, vehicle engineering) and defer V&V hiring until regulatory pressure forces it. The pattern repeats across companies and across segments. Companies that buck the pattern and hire V&V leadership in the first 30 engineering hires consistently move through certification and customer acceptance faster than companies that defer.

Certification engineering specifically is the single hardest-to-fill role in autonomous systems. The number of engineers globally who have led certification of an autonomous platform from start to entry-into-service is in the low hundreds. Companies that recruit for this role transactionally, opening a JD when they need it, typically wait 9-14 months. Companies that maintain pre-qualified certification benches fill in 8-14 weeks.

This is the section of the corpus that decays fastest without maintenance. The talent pool is small enough that individual moves change the map, and the role specifications are too specialized to reason about from generic categories. The depth here is built across years and doesn't replicate quickly. It's also where founders most often realize, late, that they should have hired earlier.