Principal Architect - AI Factory (Physical AI)

Career Area:
Technology, Digital and Data

Job Description:

Your Work Shapes the World at Caterpillar Inc.

When you join Caterpillar, you're joining a global team who cares not just about the work we do – but also about each other. We are the makers, problem solvers, and future world builders who are creating stronger, more sustainable communities. We don't just talk about progress and innovation here – we make it happen, with our customers, where we work and live. Together, we are building a better world, so we can all enjoy living in it.

Help Build the Future of Caterpillar –

At Caterpillar, technology always has a purpose, which is to solve our customers’ toughest challenges. Through Cat Technology, we are solving problems by building the intelligence layer that connects machines, data, and people to make jobsites safer, more productive, and more sustainable. By combining deep domain expertise in physical systems with software, connectivity, autonomy, and AI, we deliver solutions that work in the real world—on real jobsites, at global scale.

You’ll build and deploy against one of the most unique data foundations—over 1.6 million connected assets generating real-world data daily. These data and platform capabilities are enabling the development of AI models, edge computing architectures, and software systems that scale across fleets, products, and industries. The result will be a new generation of machines that continuously learn, improve, and deliver performance at scale.

Be Part of What’s Next in Autonomous Construction Sites

Construction autonomy is one of the most complex challenges in applied AI, and at Caterpillar, advancements in physical AI, simulation, sensing, and edge computing are turning things that once felt impossible—intelligent machines operating in dynamic jobsites—into reality.

Our connected ecosystem brings together massive volumes of high-quality data to create a foundation where engineers like you can build and deploy against.

If this work motivates you, we invite you to join our team. In these roles, you’ll work at the intersection of the physical and digital worlds. You’ll help design and deliver intelligent systems that enable machines to perceive their environment, make informed decisions, and support safer, more productive operations.

Apply today to build the new era of construction autonomy at Caterpillar.

Job Summary:

The Principal Architect for AI Factory is responsible for defining the data architecture that enables Physical AI and autonomous systems to learn, reason, and operate at scale. This role establishes the data domains, canonical data models, data products, and architectural standards that transform vast amounts of jobsite, machine, sensor, telemetry, simulation, and operational data into trusted, reusable assets for AI development and autonomy. Working at the intersection of data, AI, digital twins, and autonomy, the architect provides technical leadership for how data is organized, managed, and delivered across the enterprise, ensuring engineers and data scientists can efficiently develop, train, fine-tune, validate, and deploy next-generation AI models while accelerating innovation through scalable and reusable data products.

Bottom line, you are creating the data foundation that enables AI innovation at scale.

What You Will Do:

- Developing detailed architecture deliverables to solve business problems.

- Designing an application's technical infrastructure, such as specific databases, programming languages, utilities, and testing approaches.

- Leading the evaluation and deployment of new technologies to add or enhance existing digital technical capabilities.

- Participating in addressing business requirements for applications and collaborating with cross-functional teams to deliver digital solutions that meets business results.

What You Will Have:

- Data Architecture: Knowledge of processes, techniques and factors that affect data architecture; ability to design blueprints on how to integrate data resources for business processes and functional support.

- Analytical Thinking: Knowledge of techniques and tools that promote effective analysis; ability to determine the root cause of organizational problems and create alternative solutions that resolve these problems.

- Platform Architecture: Knowledge of technologies and methods to design processing mechanisms and roadmaps to execute business application systems; ability to design these roadmaps and deploy supportive interfaces for end-users to access related systems, in accordance with standards and processes.

- Requirements Analysis: Knowledge of tools, methods, and techniques of requirement analysis; ability to elicit, analyze and record required business functionality and non-functionality requirements to ensure the success of a system or software development project.

- Target Architecture: Knowledge of target architecture; ability to develop the IT blueprint and roadmap while aligning the architecture and processes with business strategies and objectives roadmap while aligning the architecture and processes with business strategies and objectives

Top Candidates Will Have:

- AI Factory and GPU Infrastructure: Typically 8+ years architecting distributed AI infrastructure on-premises and in the cloud: multi-node GPU clusters (DGX / HGX / GB200-class or cloud equivalents), NVLink / InfiniBand / Spectrum-X fabrics, high-throughput AI storage, and cluster sizing.

- Workload Orchestration – Deep Kubernetes for GPU workloads (operators, multi-tenant queues, quotas) plus Slurm, Run:ai / NVIDIA Mission Control, or Ray; infrastructure as code (Terraform, Ansible, Helm); hybrid cloud / on-prem.

- Distributed Training and Performance – PyTorch FSDP / TensorRT/ DeepSpeed / Megatron, NCCL collectives, parallelism strategies, profiling (Nsight), elastic and fault-tolerant training; drives goodput and cost per unit of data.

- Data Flywheel for Physical AI – Petabyte-scale multimodal sensor data: ingestion, lakehouse / object storage, distributed ETL (Spark, Ray Data, Airflow, Flyte), curation, labeling and auto-labeling, dataset versioning and lineage, synthetic data integration.

- MLOps and Model Lifecycle – Training, fine-tuning, evaluation, registry, deployment, monitoring, and governance with MLflow, Weights & Biases, Kubeflow, or similar; VLA / foundation-model post-training workflows.

- Simulation-in-the-Loop and RL Infrastructure – Synthetic data pipelines (Omniverse Replicator, Isaac Sim, NVIDIA Cosmos), large-scale closed-loop RL infrastructure (Isaac Lab, Ray RLlib), and simulation / replay-based evaluation orchestrating millions of jobs.

- NVIDIA Ecosystem and Edge Deployment – DGX, NVIDIA AI Enterprise, NeMo, NIM, Triton, TensorRT, Dynamo, CUDA, and Jetson / DRIVE-class edge compute; path from trained model to optimized, monitored inference on machines.

- Observability, Reliability, and Security – DCGM / Prometheus / Grafana / Loki, SRE practices and SLOs, FinOps for GPU cost, and security and governance controls for safety-relevant autonomy software.

- Software Engineering – Strong Python and working C++; containers, CI/CD, Bazel or monorepo tooling, cloud (AWS / Azure); sound build-versus-buy judgment.

- Architecture Leadership – Translates architecture into implementation plans, reference architectures, and standards across platform teams; partners with Cat Digital Division and NVIDIA / cloud providers on enterprise platform standards.

- Domain Exposure – ML infrastructure or data / compute platforms for AV or robotics (Applied Intuition, Waabi, Wayve, Waymo, Zoox, Tesla, NVIDIA, Aurora), or AI Factory reference architectures (NVIDIA, Dell, HPE, Siemens).