PNNL

Derek Lilienthal is a Software Engineer on the Research AI Workflows Team at Pacific Northwest National Laboratory (PNNL), where he specializes in scaling AI applications across high-performance computing (HPC) and cloud environments.

With a background that bridges both software development and research, he brings a rare end-to-end perspective — from translating early-stage research ideas into robust, production-ready applications. His recent work has centered on the full lifecycle of LLM-agent systems: architecting, implementing, and evaluating agentic pipelines with a strong emphasis on cloud deployment and scalability.

His technical expertise spans the development of Infrastructure-as-Code (IaC), scalable and stateless REST APIs (leveraging both traditional HTTP servers and MCP), and agent orchestration frameworks designed for reusability and extensibility. He also integrates observability tooling such as Langfuse to ensure reliable monitoring and evaluation of deployed systems.

Core competencies include:

• Efficient distributed training of vision-language models on multi-node, multi-GPU clusters
• Scalable cloud infrastructure deployment on AWS
• LLM application development with LlamaIndex, Langchain, and LangGraph
• Containerization and CI/CD automation using Docker and GitLab

Notable projects include the development of ChatNEPA, a custom end-to-end Agentic RAG pipeline for PermitAI; distributed LLM training on on-premises HPC systems; LLM integration tutorials for HPC documentation; and the design of multi-agent systems and sandbox environments for rigorous LLM testing.

• Master of Science in Artificial Intelligence, San Jose State University
• Bachelor of Science in Computer Science, California State University