AI Development Engineer

Job description

Career area: Engineering.

Your work shapes the world at Caterpillar. Join a global team building stronger, more sustainable communities and solving problems alongside customers where they work and live.

Role: Senior Software Engineer – Autonomous-Driving Data Loop and AI Fault-Diagnosis Platform.

Responsibilities

• Develop on-vehicle fault-event detection and automatic data-segmentation/upload tools, triggered by faults in perception, localization, planning, and control.
• Build an enterprise autonomous-driving data-loop platform to manage returned data across multiple mines, vehicles, and software versions and integrate with data-labeling platforms.
• Develop automated issue diagnosis using vision-language models (VLMs), organizing images, point clouds, trajectories, logs, and fault codes as model context to produce fault descriptions, root-cause analyses, and troubleshooting recommendations.
• Use RAG, tool calling, or agent workflows for complex diagnosis, with evidence citations and human confirmation to control hallucinations.
• Build VLM domain fine-tuning and evaluation (SFT, LoRA/QLoRA), continually incorporating field data.
• Build MLOps pipelines for one-click training, validation, export (ONNX/TensorRT), and in-vehicle deployment of perception and fusion models.
• Design platform architecture and develop core components for scalability, high availability, and observability.

Qualifications

• Bachelor’s degree or higher in computer science, software engineering, AI, automation, or related field.
• More than five years of software-platform, autonomous-driving data-platform, or AI-engineering-system development.
• Proficiency in Python and one of C++, Go, or Java; familiarity with web backends, REST APIs, database design, Docker, Linux, and CI/CD.
• VLM/multimodal-model project experience and ability to carry work from solution design through deployment. Fine-tuning methods such as SFT/LoRA/QLoRA and multimodal training/evaluation datasets. Prompt engineering, RAG, vector search, tool use, or agent workflows. VLM inference deployment/optimization, including quantization, GPU-memory optimization, and acceleration.
• Platform engineering: React/Vue/TypeScript for complex data visualization; databases, object storage, caches, and message queues; ML/deep-learning training, evaluation, and deployment.
• Preferred: autonomous driving, robotics, ADAS, or mining automation; fault diagnosis, data feedback, or corner-case mining; ROS/ROS 2, ROS Bag, MCAP; open models such as LLaVA, Qwen-VL, InternVL, CLIP; Kubernetes, GPU scheduling, vLLM, Triton, MLflow, or Kubeflow.

Posting dates: September 18–October 8, 2026. Caterpillar is an equal-opportunity employer; qualified applicants of any age are encouraged to apply. Not…