How to Build an Autonomous Vehicle Engineering Team in Singapore in 7 Steps

Singapore is building the most comprehensive autonomous vehicle ecosystem in Southeast Asia. With CETRAN at NTU providing world-class testing infrastructure, the Land Transport Authority issuing trial licences for on-road autonomous vehicle operations, and the Ministry of Transport's CVES framework delivering vehicle-to-infrastructure communication across the island, the regulatory and physical infrastructure is in place. The constraint is talent.

The global AV industry is in a paradoxical moment: companies like Uber, Cruise, and Motional have displaced thousands of engineers with production autonomous driving experience, while Singapore's AV programmes are expanding faster than local universities can produce qualified graduates. This guide provides a structured 7-step approach for Singapore employers who want to build an AV engineering team that can develop, test, and deploy autonomous vehicles on Singapore's roads.

Whether you are an established transport operator (Grab, ComfortDelGro, SMRT), an AV startup based in one-north or Jurong Innovation District, a logistics company automating Tuas Port operations, or an international AV company opening a Singapore R&D centre, this framework applies. The steps are sequenced deliberately: each one depends on the previous step's output.

Step 1: Define Your AV Mission and Role Architecture

Before you write a single job description, answer one question: what will your autonomous vehicles do in Singapore, and where? The answer determines which engineering disciplines you need and in what proportions.

The AV landscape in Singapore spans four distinct application domains, each requiring a different team composition:

• Autonomous shuttles (low-speed, fixed route): Perception requirements are moderate because routes are predefined and speeds are under 25 km/h. Planning is constrained but must handle pedestrian interactions at bus stops and crossing points. Team emphasis: safety engineering and V2X integration. Example deployment: one-north shuttle between Buona Vista MRT and research parks.
• Autonomous taxis (mixed traffic, dynamic routing): The most technically demanding application. Full perception stack (LiDAR, camera, radar fusion), unconstrained motion planning, real-time dispatch integration, and passenger safety systems. Team emphasis: perception, planning, and systems architecture. Example: ComfortDelGro autonomous taxi trials in the CBD.
• Autonomous freight and logistics (port operations, last-mile delivery): Heavy emphasis on precision localisation in structured environments (port yards, warehouse districts), heavy vehicle dynamics, and cargo safety. Team emphasis: controls engineering and industrial safety. Example: Tuas Port autonomous container movement.
• Autonomous buses (high-capacity, scheduled routes): Combines elements of shuttles (fixed route) and taxis (mixed traffic, higher speeds). Additional requirements for passenger boarding/alighting systems and accessibility compliance. Team emphasis: safety engineering and public-facing HMI. Example: SMRT autonomous bus on the Jurong Innovation District route.

Once you have defined your application domain, map it to a role architecture. A minimum viable AV team for any of these domains requires these core roles:

• AV Systems Architect / Tech Lead (1): Defines the overall system architecture, makes technology selection decisions, and ensures integration across perception, planning, controls, and safety subsystems. Requires 8+ years with at least 4 in production AV systems.
• Perception Engineers (2-3): Build and maintain the sensor fusion pipeline — LiDAR point cloud processing, camera-based object detection (pedestrians, vehicles, cyclists, road markings), radar tracking, and multi-sensor temporal fusion.
• Motion Planning and Controls Engineers (2-3): Design algorithms for path planning, behaviour planning (lane changes, intersection traversal, yielding), and vehicle controls (steering, throttle, braking commands).
• Simulation and Testing Engineer (1-2): Build and maintain the simulation environment, design test scenarios, manage regression testing, and coordinate CETRAN physical testing campaigns.
• Functional Safety Engineer (1): Conducts hazard analysis and risk assessment (HARA), defines safety requirements per ISO 26262 and ISO 21448 SOTIF, verifies redundancy architectures, and validates fail-safe behaviour.
• V2X / Connectivity Engineer (1): Integrates with Singapore's CVES infrastructure — DSRC roadside units, C-V2X base stations, traffic signal phase data, and vehicle-to-vehicle communication protocols.

Total: 8 to 12 engineers for a team that can develop, test, and iterate on a production AV stack. Scale up as your programme progresses from prototype to trial licence to commercial deployment.

Step 2: Establish Regulatory and Testing Partnerships

Singapore's AV regulatory environment is collaborative, not adversarial. The government actively supports AV development through three institutional pillars that your team must engage with before putting any vehicle on a public road:

CETRAN at NTU

The Centre of Excellence for Testing and Research of Autonomous Vehicles (CETRAN), located at Nanyang Technological University in Jurong, is Singapore's primary AV testing facility. Its 1.8-hectare test circuit includes:

• Multi-lane expressway simulation with merging and diverging ramps
• Signalised intersections with configurable traffic light phasing
• Bus stop and taxi stand interaction zones
• Pedestrian crossing scenarios with varied signalisation
• HDB estate driving environment with narrow roads and parked vehicles
• Artificial rain system simulating Singapore's tropical downpours (up to 200mm/hour intensity)
• Variable lighting conditions (tropical sun glare, night operations, tunnel transitions)

CETRAN offers tiered testing programmes. For early-stage companies, start with the component validation programme to test individual perception or planning modules. For teams ready for integrated vehicle testing, the scenario-based assessment programme evaluates your full AV stack against standardised Singapore road scenarios. CETRAN testing results are accepted by the LTA as part of the trial licence application process.

LTA Trial Licences

The Land Transport Authority issues autonomous vehicle trial licences that permit on-road testing in designated zones. The application requires a safety case document, evidence of simulation testing coverage, CETRAN test results, insurance coverage (minimum SGD 5 million third-party liability), and a detailed operations plan including remote monitoring and safety driver protocols.

Current trial zones include one-north (Buona Vista area), Jurong Innovation District (JID, connecting JTC facilities and NTU campus), and Punggol Digital District (PDD, smart district for last-mile mobility). Each zone has specific speed limits, operating hour restrictions, and weather suspension criteria.

MoT CVES Framework

The Ministry of Transport's Connected Vehicle Enabling System (CVES) provides the V2I communication infrastructure that gives your autonomous vehicles real-time traffic data. Engaging with the CVES team early — ideally during your team formation phase — allows you to design your AV software stack to integrate CVES data from the ground up rather than retrofitting it later.

Step 3: Build Your AV Tech Stack Foundation

Your tech stack decisions constrain which engineers you can hire. An Autoware-based stack requires engineers who know ROS2 and C++. A proprietary stack gives you more architectural freedom but requires more senior engineers who can build from scratch. Here is the decision framework:

Open-Source Foundation (Recommended for Most Singapore Teams)

The Autoware open-source autonomous driving platform, built on ROS2, provides a production-grade foundation for perception, planning, and controls. Singapore has an active Autoware Foundation chapter, and NTU researchers contribute to the project. Starting with Autoware gives you:

• A working AV stack on day one that your team can modify and extend
• Access to a global community of AV engineers who contribute perception models, planning algorithms, and simulation tools
• Compatibility with standard sensor hardware (Velodyne, Ouster, Hesai LiDAR; FLIR cameras; Continental radar)
• CETRAN familiarity — CETRAN's engineers are experienced with Autoware-based vehicles, which simplifies the testing process

For simulation, CARLA (open-source, Unreal Engine-based) is the most common choice in Singapore's AV ecosystem. It supports sensor simulation (LiDAR, camera, radar), traffic participant simulation, and weather effects including the tropical rain scenarios critical for Singapore operations. LGSVL Simulator (if using its open-source fork) and NVIDIA DRIVE Sim (for teams with NVIDIA hardware partnerships) are alternatives.

Compute Platform

For in-vehicle compute, NVIDIA DRIVE Orin remains the standard for production AV deployments in 2026, offering 254 TOPS of AI compute in a 45W power envelope. For development and testing, NVIDIA Jetson AGX provides a cost-effective alternative. Ensure your team has software engineers who can optimise perception models for the target compute platform using TensorRT and CUDA.

Step 4: Source AV Talent From Global Pools and Local Labs

AV talent sourcing in Singapore requires a multi-channel strategy. No single source will fill your team. Here are the five channels, ranked by quality and speed:

Channel 1: Global AV Layoff Pools (Highest Quality, Time-Sensitive)

The 2025-2026 wave of AV industry restructuring has displaced engineers with production autonomous driving experience. Key pools include:

• Uber (September 2026): 3,300 employees laid off, including an estimated 300-450 AV-adjacent engineers across perception, planning, simulation, and safety.
• Cruise (GM): Suspended operations in late 2023 after a pedestrian incident; many engineers have been on the market for 18+ months and are actively looking for non-US opportunities.
• Motional (Hyundai-Aptiv JV): Scaled back operations in 2024-2025, displacing perception and planning engineers with Ioniq 5-based AV experience.
• Argo AI alumni: Ford's AV unit shut down in 2022; experienced engineers who did not join Waymo or Aurora are still available.

The window for these candidates is 60 to 90 days from each layoff announcement. After that, the most experienced engineers have accepted offers from Waymo, Aurora, Apple, or Chinese AV companies (Pony.ai, WeRide, AutoX). Act fast.

Channel 2: NTU and NUS Graduates

NTU's School of Mechanical and Aerospace Engineering and NUS's Department of Mechanical Engineering produce approximately 50 to 80 graduates per year with robotics, autonomous systems, or computer vision specialisations. These are junior candidates (0-2 years experience) who require mentoring but understand Singapore's road environment and can contribute to CETRAN testing campaigns immediately. Build relationships with NTU's Intelligent Transportation Systems Lab and NUS's Advanced Robotics Centre to access students before graduation.

Channel 3: A*STAR Researchers

The Agency for Science, Technology and Research (A*STAR) operates the Institute for Infocomm Research (I2R), which has active programmes in sensor technology, computer vision, and embedded AI — all directly applicable to AV development. A*STAR researchers often seek industry transitions after completing 3-5 year research programmes. They bring deep technical expertise and publication track records, though they may need onboarding into production engineering practices.

Channel 4: International AV Labs in Singapore

Hyundai, BMW, Bosch, and Continental all operate R&D facilities in Singapore with autonomous vehicle programmes. Engineers at these facilities may be open to joining Singapore-based AV startups or operators that offer more technical ownership and faster deployment timelines. Approach with respect for non-compete and non-solicitation agreements.

Channel 5: Tech.Pass for Senior International Hires

For principal engineers, AV architects, and functional safety leads, Singapore's Tech.Pass programme provides a personalised work pass for individuals earning at least SGD 20,000 per month. Tech.Pass holders can start or join any company without an employer-tied visa, making it attractive for senior candidates who want flexibility. Processing takes 4 to 8 weeks.

Step 5: Design AV-Specific Technical Interviews

Standard software engineering interviews (LeetCode, system design, behavioural) do not adequately assess AV engineering competence. An AV interview process must evaluate domain-specific reasoning about safety-critical systems. Here is a four-stage framework:

Stage 1: Technical Screen (45 min, remote)

A focused conversation on the candidate's AV experience. For perception engineers, ask them to walk through their sensor fusion pipeline architecture, explain how they handle LiDAR-camera calibration drift, and describe their approach to detecting occluded pedestrians. For planning engineers, ask about their behaviour planning state machine, how they handle unprotected left turns at intersections, and their approach to emergency stop decisions. The goal is to separate candidates with real production AV experience from those with only simulation or academic experience.

Stage 2: Domain Design Exercise (90 min, remote or on-site)

Present a Singapore-specific AV scenario and ask the candidate to design a solution. Examples:

• Perception: Design a sensor configuration for an autonomous shuttle operating in the one-north precinct. The route includes a covered walkway section (no GPS), a rain-exposed road crossing, and a shared pedestrian-vehicle zone. Specify sensor types, placement, and your fusion strategy.
• Planning: Design a behaviour planner for an autonomous taxi navigating the Shenton Way CBD during evening rush hour. The vehicle must handle double-parked delivery vans, jaywalking pedestrians, buses pulling into and out of bus stops, and construction zones with temporary lane markings. Define your state machine and priority hierarchy.
• Safety: Conduct a preliminary hazard analysis for an autonomous bus operating the Jurong Innovation District route. Identify the top 5 hazards, assess their severity and exposure, and propose safety requirements with ASIL classifications per ISO 26262.

Stage 3: Code and Simulation Assessment (2-3 hours, take-home or on-site)

A hands-on exercise using your actual (or representative) tech stack. For perception candidates, provide a labelled point cloud dataset from a Singapore road environment and ask them to implement a 3D object detection pipeline. For planning candidates, provide a simulation scenario in CARLA and ask them to implement a planner that navigates the scenario safely. For simulation engineers, provide a set of edge cases and ask them to design the simulation test suite.

Stage 4: Safety and Ethics Discussion (45 min, on-site)

AV engineering involves decisions that directly affect human safety. This stage assesses the candidate's approach to safety-critical decision-making, their understanding of the ethical dimensions of autonomous mobility, and their ability to communicate safety trade-offs to non-technical stakeholders. Ask about scenarios where they had to choose between system performance and safety margin, how they handled a near-miss event in testing, and how they would explain their system's limitations to a regulator.

Step 6: Structure Competitive Compensation Packages

AV engineers are among the most highly compensated engineering specialisations globally. Singapore compensation is lower than Silicon Valley in absolute terms but competitive on a take-home basis. Here is the 2026 compensation framework:

For international candidates relocating from the US, frame your offer around take-home compensation. A SGD 22,000/month package in Singapore (SGD 264,000/year) results in approximately SGD 220,000 after tax at Singapore's effective rate. A comparable USD 250,000 base in San Francisco results in approximately USD 155,000 after federal and California state taxes. The Singapore take-home is effectively higher, before accounting for Singapore's lower cost of living in housing (outside of prime districts), transport (world-class public transit), and healthcare.

Beyond base salary, competitive packages for AV engineers should include:

• Equity or phantom equity: AV startups should offer 0.1% to 0.5% for senior engineers, 0.5% to 1.5% for architects. Established companies should provide RSU-equivalent mechanisms.
• Relocation package: SGD 10,000 to 25,000 for international hires, covering flight, temporary housing, and settling-in costs.
• Professional development: Budget for conference attendance (IEEE ITSC, IV Symposium, ROSCon), safety certification training (ISO 26262 courses), and continued education.
• Hardware access: AV engineers expect access to sensor hardware, test vehicles, and compute infrastructure. Budget SGD 200,000 to 500,000 for initial hardware setup.

Step 7: Onboard With a 90-Day AV Integration Playbook

AV engineers, particularly those relocating from US AV programmes, need structured onboarding that goes beyond standard corporate orientation. Your 90-day integration playbook should cover:

Week 1-2: Singapore Context Immersion

• Singapore road rules and driving culture: Left-hand traffic, ERP gantry behaviour, HDB estate driving norms, bus lane rules, and school zone regulations. International engineers need hands-on driving experience in Singapore to understand the environment their algorithms must navigate.
• CVES infrastructure briefing: Orientation on Singapore's V2I communication infrastructure, data formats (SPaT, MAP messages), and integration APIs. Arranged through MoT's CVES programme office.
• Regulatory framework overview: LTA trial licence requirements, safety case documentation standards, incident reporting protocols, and data retention obligations under Singapore law (PDPA compliance for AV sensor data that captures identifiable individuals).

Week 3-4: CETRAN Orientation

• Test track familiarisation: On-site sessions at CETRAN to understand the test circuit layout, available scenarios, and testing protocols.
• Simulation-to-real gap analysis: Compare your simulation scenarios with CETRAN's physical test results to identify where your simulation diverges from Singapore road reality. Tropical rain, equatorial sun glare, and high-humidity sensor degradation are common gaps.
• Safety validation workflow: Define the end-to-end process from code change to simulation regression to CETRAN physical test to on-road trial. Every engineer should understand this pipeline.

Week 5-8: First Contributions

• Perception engineers: First object detection model trained on Singapore-specific data (buses, motorcycles, PMDs, construction barriers). Evaluate performance in tropical rain degradation scenarios.
• Planning engineers: First planning module integrated into the simulation pipeline. Run regression tests against baseline Singapore road scenarios.
• Safety engineers: Complete preliminary hazard analysis for the target deployment zone. Begin drafting the safety case document for LTA trial licence application.

Week 9-12: First CETRAN Test Campaign

• Schedule and execute the first physical test campaign at CETRAN. Use results to validate simulation fidelity and identify sim-to-real gaps.
• Conduct the first team retrospective: what worked, what the simulation missed, where the tech stack needs modification.
• Produce the first iteration of your LTA trial licence safety case document, incorporating CETRAN test evidence.

By the end of week 12, your team should have a functioning AV software stack in simulation, initial CETRAN test results, a draft safety case, and a clear roadmap for on-road trial licence application. This timeline is aggressive but achievable for teams that start with an Autoware foundation and hire engineers with production AV experience.

Frequently Asked Questions