ADAS Architect - Simulation, Data & Verification Platform

ADAS Architect - Simulation, Data & Verification Platform

About the team

You will join a focused team of domain experts building a greenfield, in-house L2+ ADAS software stack for Indian road conditions, covering four-wheelers and two-wheelers. The team designs and owns the complete architecture, algorithms, and tooling, from first principles through to deployment on automotive-grade ECUs. Engineers here shape foundational technical decisions, work directly with experts across perception, fusion, driving functions, motion control, and simulation, and see their work carried from concept to vehicle.

Role purpose

The Simulation, Data & Verification Platform Architect defines and builds the engineering platform on which every ADAS algorithm in the organization is developed, verified, and released. This is not the integration or configuration of an existing framework: the role creates, from the ground up, a proprietary simulation engine and developer application, the automated verification and reporting system that runs on it, and the data platform that feeds both, from recorded vehicle data through labelling to cloud-scale execution.

The platform is the primary evidence base for every function, before and after vehicles are available. Its correctness, determinism, and traceability directly determine the credibility of every result the organization reports. The role calls for a senior architect with deep software engineering skill, strong automotive domain understanding, and a record of building platforms that entire engineering teams depend on.

Key responsibilities

Simulation engine and developer application

• Architect and build a proprietary simulation engine that executes any algorithm component delivered as a host-agnostic C-API library, discovering the component's ports and configuration at run time rather than through hand-written integration.
• Define the execution model: closed-loop simulation, deterministic resimulation against recorded data, and pure replay as distinct, well-defined modes; multi-rate scheduling; and multi-component pipelines in which any stage can run live, stubbed, or injected from a recording.
• Design and build the interactive developer application: timeline and transport control, 2D/3D scene view, signal plots, port and message inspection, layer-based visualization contributed by each component as plug-ins, and side-by-side comparison of runs.
• Guarantee traceability: every value displayed or reported is traceable to the recording, message, build, and configuration that produced it, and the headless command-line interface produces answers identical to the GUI.
• Establish determinism as a measured property: bit-exact resimulation within and across processes, and tracked, explained differences across builds and machines.
• Provide process isolation, timeouts, and crash capture so that a failing component never takes down a run or a campaign.
• Integrate external simulators and scenario tools (e.g., CarMaker, CARLA, esmini) through codecs and adapters, without coupling any algorithm component to a specific tool.

Automated verification and reporting framework

• Design and build, from zero, the automated test framework that runs scenario campaigns on the simulation engine: scenario definitions with machine-checkable acceptance criteria, parameter sweeps, and NCAP-based and India-specific scenario libraries.
• Build automated reporting for every campaign: per-scenario verdicts with evidence, function KPIs, coverage against requirements, trend and regression tracking, and drill-down from any failed case to the exact moment in the recording.
• Implement regression gates in CI that block a change when a KPI, scenario outcome, or determinism check regresses.
• Link verification results to requirements so that traceability and coverage reports are generated automatically, never maintained by hand.

Data platform: recordings, storage, labelling, and cloud execution

• Define the recording data model and formats (e.g., MCAP, MDF4, rosbag), including embedded schemas and provenance (vehicle, sensor set, software build, configuration, and scenario identity).
• Architect storage and data management for large volumes of vehicle recordings: ingestion from test vehicles, indexing, metadata and search, retention, dataset versioning, and access control.
• Establish labelling and annotation pipelines for ground truth: tool selection or development, auto-labelling, quality control, annotation specifications, and management of external annotation vendors.
• Build scenario extraction and mining from recorded data, turning field drives into reusable, parameterized test scenarios.
• Architect cloud and cluster execution of resimulation and scenario campaigns at scale (thousands of runs), including job orchestration, containerization, cost control, and reproducibility of every run.

Technical leadership

• Own the technical roadmap of the platform and make build-versus-buy decisions with clear justification.
• Set engineering standards for the platform: code quality, testing, documentation, and backward compatibility of interfaces.
• Work with algorithm, perception, test, and integration teams as internal customers, and with the ECU-in-the-Loop Test & Release Architect so that SIL, HIL, and vehicle results share one reporting system.
• Mentor engineers and, as the platform grows, build and lead the team around it.

Required qualifications

• Bachelor's or Master's degree (PhD a plus) in Computer Science, Electrical/Electronics, Mechatronics, Robotics, or a related discipline.
• 7-15 years of software engineering experience, including at least 5 years architecting simulation, SIL/HIL, data, or verification platforms for ADAS, autonomous driving, or robotics.
• Demonstrated design and delivery, from scratch, of a simulation or resimulation platform adopted as the primary development and regression tool of an engineering organization.
• Expert-level C++ (C++17) and Python: C ABIs, dynamic library loading, plug-in architectures, cross-language bindings, and performance-critical, multi-threaded code on Windows and Linux.
• Experience building desktop engineering applications with rich visualization (e.g., Qt, Dear ImGui, OpenGL/Vulkan, or equivalent).
• Deep knowledge of recorded-data formats and large-scale handling of automotive sensor data (camera, radar, lidar, vehicle bus).
• Hands-on experience with ground-truth labelling and annotation workflows for ADAS data, including quality metrics.
• Proven experience with cloud or cluster execution (e.g., AWS, Azure, GCP, Kubernetes, containerized batch workloads) for simulation or data processing at scale.
• Strong knowledge of build, packaging, and CI (CMake, Conan or equivalent, and modern CI systems).
• Experience with schema-driven interfaces (e.g., Protocol Buffers) and long-term interface versioning.
• Working knowledge of ADAS function behaviour, NCAP test protocols, and scenario-based testing (ASAM OpenSCENARIO / OpenDRIVE).

Preferred qualifications

• Experience building a scenario-mining or data-driven test-generation pipeline from fleet or test-vehicle data.
• Experience with auto-labelling using offline perception models.
• Experience with verification in regulated contexts (ISO 26262, ISO 21448 SOTIF, ASPICE) and with tool qualification.
• Contributions to open-source simulation, robotics, or data tooling.
• Experience leading or growing a platform team.
• Advanced use of AI-assisted software development tools to accelerate platform development.

Key competencies

• Ability to explain complex trade-offs clearly to engineers and leadership to drive technical decisions.
• Architectural judgement: designs that stay simple as components, functions, and data volumes grow by orders of magnitude.
• Uncompromising rigor in determinism, traceability, and reproducibility: a result that cannot be traced is not accepted as a result.
• A product mindset toward internal tools: engineers adopt the platform because it is the fastest way to work, not because it is mandated.
• Ownership from first design to organization-wide daily use, without reliance on handoffs.
• Ability to build new systems where architecture and processes are still being defined, and to make them the reference for the organization.
• Clear technical communication with engineers, architects, and management.

L&T Company Overview / About Us:

L&T is a USD 30 billion Indian multinational engaged in EPC Projects, Hi-Tech Manufacturing, and Services, operating across multiple geographies. A strong, customer-focussed approach and the constant quest for top-class quality have enabled L&T to attain and sustain leadership in its major lines of business for eight decades.

Given L&T’s capability in strategic electronics and leveraging its foray into semiconductor design and well-established capabilities in L&T Technology Services, the business is expanding its offerings to cover industrial electronic modules, products and systems in critical sectors of Power Electronics, Automotive Electronics, Industrial Robotics & Automation, Communication Platforms and Electronic Systems Design & Manufacturing (ESDM). The focus products under Automotive Electronics are ADAS, Electrification & Connectivity related products.