HIRING & TALENT Update – Autonomous Driving London

For Hiring Managers & Engineers · Autonomous Driving Recruitment

The Future of Autonomous Driving: London Is the Test Track

Which engineering skills will decide who wins, what the London trials are teaching every major city and where engineering and technical hiring demand is heading into 2027.

By Phil Bowers, Tiro Associates  |  2 October 2026  |  12 minute read

London is now the most closely watched test of whether autonomy can work in an old, dense and unforgiving city. Wayve is carrying paying passengers, Waymo and Baidu are learning the streets and the rulebook is still being written. The winners will not be the companies with the most cars. They will be the ones that hire the scarce engineers who turn street data into a safety case a regulator will sign.

1. London is the test track

Three programmes, one city

On 3 September 2026 Uber and Wayve began the first supervised autonomous rides for paying passengers in the UK. Riders booking UberX, Uber Comfort or Uber Electric can be matched with a Wayve car at no extra cost, with a licensed safety driver behind the wheel. The fleet started at around 15 vehicles and sits inside a wider Wayve and Uber plan covering 12 international markets.

Wayve is not alone. Waymo has been gathering driving data across 19 London boroughs and has said its London service will be backed by hundreds of UK jobs and tens of millions of pounds of local investment, with fleet operations run by a partner, Moove. Baidu began testing its sixth generation Apollo Go RT6 in the borough of Brent on 28 July 2026 with Freenow by Lyft. It is also working with Uber and is aiming for public riders in 2027.

Why London matters so much: Wayve’s leadership has pointed out that the city has more than ten times the cyclist and pedestrian density of San Francisco. Roundabouts, uncontrolled crossings, double parked vans, buses pulling out and lanes narrower than any Sunbelt city all arrive at once. Government has framed the wider prize as 38,000 jobs and £42bn for the economy. If autonomy can be proven here, the safety argument travels to almost any major city.

2. Near misses: the data that decides who gets a permit

Converting street data into safety evidence

The London trials have already produced the kind of moments regulators care about most. A cyclist approached a roundabout and the safety driver took manual control and braked; when asked how often humans step in, the operator said it did not hold that figure and did not use it as a primary metric. A test car under manual control drove into a police cordon in Harlesden and the driver was suspended, a reminder that the human layer is part of the safety system too. An early press ride in a Wayve car became stuck in a narrow passage of the kind London drivers handle by instinct.

None of these were collisions. All of them are evidence. Government opened its consultation on the statutory Statement of Safety Principles on 17 June 2026, with draft principles that include predicting and responding to hazards proactively and interacting safely with vulnerable road users and emergency services. Respondents to the earlier call for evidence asked government to define what counts as a near miss, to report near misses per distance travelled and even to build a UK wide incident and near miss database. The final statement is due before Parliament in 2027.

That turns near miss handling into an engineering pipeline: capture every intervention and hesitation, triage it, reconstruct the scenario, find the root cause, regress the fix in simulation, update the safety case and report it in a form a regulator accepts. The data already exists in terabytes. The bottleneck is the small number of people who can run that loop with the rigour of an air accident investigator and the speed of a software team.

3. One city, one rulebook

Why autonomy has to be delivered country by country

The Automated Vehicles Act 2024 created the legal framework, but the Act alone does not authorise a commercial driverless service. The Automated Passenger Services permit regulations took effect on 15 May 2026, alongside a pilot scheme that for the first time allows driverless operation without a safety driver. An operator needs vehicle approval, a Vehicle Special Order, listing as an automated vehicle and an APS permit issued by the DVSA. In London, no permit can be issued without the consent of Transport for London, whose Commissioner said earlier this year that no autonomous vehicle operating in the UK yet met the standard to carry passengers in the capital. By late August no APS permit had been issued to any operator.

Every major city repeats this with its own variations. Germany legislated for Level 4 in 2021 and in July 2026 its federal motor transport authority granted the first nationwide authorisation for Level 4 testing on urban roads. In June 2026, 18 EU member states signed a joint declaration on cross border autonomous testbeds, the Commission plans to bring the UN driverless regulation into EU vehicle law in the fourth quarter and the revised Product Liability Directive, which reaches software, must be in national law by 9 December 2026. The US still has no comprehensive federal law and runs on a state by state patchwork. China operates at scale: Apollo Go reported 22 million rides by April 2026, including 3.2 million fully driverless rides in the first quarter alone. In July it secured Hong Kong’s first fully driverless test permit in a right hand drive system.

For employers, the lesson is that one safety case now has to be translated into several legal frameworks, several local licensing authorities and several data regimes. A Chinese developed system on London streets also raises data governance questions that a UK regulator will expect named, competent people to answer. Homologation, regulatory and cyber engineers are no longer a support function. They are on the critical path to launch.

4. Who is hiring and what demand looks like

Wayve, Waymo, Baidu and Mobileye compared

We reviewed public hiring activity across the main programmes active in London and Europe. The picture is not one of uniform growth. It is a market splitting into three waves of demand, with the same scarce profiles wanted by everyone.

Wave one, AI and software build. Still the largest volume of roles and still intense. Wayve’s growth is the clearest example, with depth in embedded platforms, model release and onboard test automation that shows a company moving from research into product. Job trackers counted anywhere from about 90 to over 300 open Wayve roles in September depending on method, which is itself a sign of how fast the picture moves.

Wave two, safety assurance and regulation. Peaking now and running into 2027 as the Statement of Safety Principles, the APS permits and the EU rules land. These are the roles that gate a launch. The pool of engineers with production safety case experience is measured in hundreds in the UK, not thousands.

Wave three, operations. Waymo’s London postings are a leading indicator: operations control, disruption management, vehicle recovery and charging infrastructure. Mobileye’s move from supplier to fleet owner shows the same shift. Running a driverless fleet in a city is its own engineering discipline. Almost nobody has done it in Europe yet.

5. The six skill sets autonomy needs

What the role covers, why it is hard to fill, how tiro find them

AI and machine learning

What it covers: end to end driving models, vision and language foundation models, training at fleet scale, evaluation and the release process that decides which model is allowed on the road.

Why it is hard to fill: London’s AV companies compete with every frontier AI lab in the city for the same people. Driving also adds real time constraints and safety evidence that most ML engineers have never had to produce.

How we find them: we map engineers who have shipped models into a regulated or physical product, in robotics, medical imaging or defence, rather than chasing research titles.

Embedded and vehicle software

What it covers: safety critical C++ and increasingly Rust, real time operating systems, middleware, onboard compute, software update pipelines, onboard test automation and reliability engineering for vehicle software.

Why it is hard to fill: the robotaxi players are pulling from the same OEM and tier one pool. Principal level packages have moved well beyond traditional automotive. One published US range for a principal embedded vehicle software role runs from $348k to $458k.

How we find them: we look beyond automotive to aerospace flight software, rail signalling and defence, where safety critical discipline already exists.

Sensors and perception

What it covers: camera, LiDAR, radar and imaging radar, calibration, sensor fusion, 3D scene reconstruction and the handling of degraded sensors in rain, glare and dirt.

Why it is hard to fill: architectures are diverging. Wayve leads with cameras and AI, Waymo with a multi sensor suite, so experience does not transfer cleanly between them. London’s occlusions, a cyclist hidden behind a bus or an HGV, are the hardest perception test going.

How we find them: we map by sensor modality and by architecture, not by job title, so a client gets people who have solved their exact problem.

Systems engineering and safety assurance

What it covers: ISO 26262 functional safety, ISO 21448 SOTIF, UL 4600 safety cases for autonomy, hazard analysis, requirements traceability and verification and validation.

Why it is hard to fill: as we said earlier this year, senior engineers with production level system safety experience take quarters, not weeks, to secure. The new UK permits make this the gate to launch, so demand is now arriving all at once.

How we find them: we work across sectors, drawing on aerospace software assurance, rail safety cases and nuclear, where engineers already defend safety arguments in front of a regulator.

Safety data, simulation and incident forensics

What it covers: near miss and disengagement triage, scenario reconstruction, simulation and scenario libraries, leading safety metrics and forensic engineering after an event.

Why it is hard to fill: it is the newest discipline in the stack. Few people have done it at fleet scale and regulators have not yet defined a near miss, so it needs judgement as well as data skills.

How we find them: we profile engineers from aviation safety investigation, motorsport data engineering and rail incident analysis, people already trained to turn telemetry into findings.

Regulation, homologation and AV operations

What it covers: type approval, the UN driverless regulation, UN R155 and R156 for cyber security and software updates, APS permits and local authority engagement, plus remote assistance, fleet operations, depot engineering and vehicle recovery.

Why it is hard to fill: the legal regime is brand new, every city adds another layer and very few people anywhere have run a driverless fleet on European roads.

How we find them: we map OEM type approval teams, rail and aviation regulatory specialists and urban fleet operators who already work under licence and scrutiny.

6. Vacancy risk: why some roles cannot wait

Assess, map, headhunt

The 2027 calendar is crowded. The Statement of Safety Principles goes before Parliament, Baidu and Freenow aim to open to public riders, MOIA is targeting driverless approval in Europe and Mobileye launches its own fleet. Senior engineers commonly carry three month notice periods, often six at principal level. A safety case lead who is not identified this quarter will not be in seat in time.

That is why we start with a vacancy risk assessment. We score each critical role on four things: how scarce the skill really is, how many named competitors in the same city are hunting the same profile, whether the role gates a permit, a safety case or a launch date and how exposed the programme is if it stays empty for six months. The output tells a leadership team which roles to open now, which to headhunt and which the in house team can fill through normal channels.

Then comes strategic mapping. We map where the skill actually sits, across London, Oxford, Cambridge, the Midlands automotive corridor and adjacent sectors such as aerospace, defence, rail and motorsport. We identify the specific capability gap being closed rather than a generic title. Finally, targeted headhunting: a direct, confidential approach to settled engineers who are not looking and will never see an advert.

It is the Tiro Way applied to autonomy:

We headhunt. We go directly to the people who fit, wherever they currently sit.

We work with our client, not their competitors. Once engaged, we do not run the same brief for a rival, which matters in a city where four programmes want the same engineers.

We plan a strategy, then execute it. The map that finds the candidate is the same plan that positions the client as the employer of choice.

We pick and choose our target candidates. Every person presented has been matched against the real gap.

We manage strategically to fill the key gaps. The hire should close the capability the team is missing.

We do not rely on hope, we manage into position. Search, approach, interview, offer, notice and start date are all actively managed.

We build strong careers for candidates and secure the best for clients. The best people join opportunities that are genuinely good for their careers.

The hard 20%. Most AV hiring runs well through an in house team. The roles that do not are the complex or unplanned ones: the safety lead who resigns three months before a permit submission, the perception gap that appears after a near miss review, the principal role that has eaten half a year of internal effort. Those candidates are not active. Reaching them takes a proper search and work outside normal hours, which is exactly what we are built for. It also frees the in house team to keep the standard application route moving for everything else.

7. For engineers: a confidential career review

A career fit conversation, not a spec fit one

If you work in safety assurance, embedded software, perception, simulation or vehicle operations, in automotive or in aerospace, rail, defence or motorsport, autonomy is probably closer to your skill set than you think. It may also not be right for you. That is worth knowing too.

We offer a confidential career review. No vacancy required and no obligation, just an honest conversation about where you sit in the market, what autonomy employers actually value and whether a move would build the career you want. We weigh it fairly against what is still good about your current role, not just sell you on leaving.

Working with Tiro

For hiring teams: if a safety case, permit or launch date depends on a role you have not filled, talk to us about a vacancy risk assessment and a strategic map of where that skill sits before your competitors reach it. Adding precision headhunting to the inhouse team can bring a real competitive edge to your talent attraction

If
For hiring advice and strategy, click here to arrange a short call Phil

Prefer email or phone? recruit@tiroassociates.com  |  Phone 44 1277 354469

A human solution for a human problem.

Tiro Associates. Autonomous driving recruitment. recruit@tiroassociates.com. Phone 44 1277 354469.