Postdoctoral Research Associate, Agentic Workflows

ORNL • Oak Ridge, TN

4d

About The Position

The Oak Ridge National Laboratory (ORNL) is seeking a dynamic Research Associate to focus on innovations in AI-integrated workflow architectures that span the Edge, Cloud, and HPC continuum. This position centers on advancing intelligent workflows that enable seamless processes in autonomous discovery, complex data integration, workflow provenance, and interactive reasoning using cutting-edge AI agents and toolkits. Your contributions will accelerate cross-disciplinary scientific progress across climate modeling, computational chemistry, additive manufacturing, and other domains by designing scalable, transparent, and adaptable workflow systems. This role offers an opportunity to work with ORNL’s leadership-class computational resources, experimental facilities, and collaborative platforms like Flowcept, CrewAI, INTERSECT, and the S3M Facility API, pushing the boundaries in how scientists interact with real-time provenance data and AI-assisted experiments. Furthermore, these advancements in intelligent AI workflows and agentic systems at ORNL may play a significant role in contributing to the 2025 Genesis Mission, towards its goals of accelerating scientific discovery through cutting-edge AI integration. Focus Areas: Modular Workflow Design : Improve scalability, portability, and reliability of loosely coupled workflows for large-scale experiments and computational workloads across diverse infrastructures. Interactive AI Agents for Scientific Workflows : Advance LLM-powered agents that interpret natural language queries, generate runtime provenance insights, reason over metadata, and steer workflows adaptively based on scientific goals. Cross-Facility Orchestration : Streamline interaction between experimental facilities, HPC systems, and edge/cloud platforms, enabling transparent data movement, multi-agent collaboration, and workflow modularization. Dynamic Workflow Schema Development : Design rich, metadata-driven dataflow schemas for tracking provenance, enabling efficient multi-step reasoning and integration across scientific domains. Agentic AI Integration with Digital Twins: Advance the use of workflow-integrated agentic AI systems in conjunction with digital twins to enable autonomous, feedback-driven experimentation and adaptive workflow reasoning. Integration of HPC + AI + QC Workflows : Develop innovative hybrid workflows that integrate High-Performance Computing (HPC), Artificial Intelligence (AI), and Quantum Computing (QC) paradigms to address next-generation scientific challenges. Workflow Benchmarking and Optimization : Design and implement robust benchmarking frameworks to analyze and optimize the performance of AI-powered workflows and ensure scalability, traceability, and efficiency across diverse infrastructures.

Requirements

• Ph.D. in Computer Science, Data Science, Computational Science, or a relevant domain discipline (completed within the last 5 years or nearing completion).
• Experience with scientific workflows, distributed systems, or AI agent development, particularly integrating LLMs or autonomous tools within complex pipelines.
• Proficiency in modern AI frameworks and tools (e.g., PyTorch, TensorFlow, LangChain, MCP SDKs) and programming languages (Python, C++).
• Experience with provenance systems (e.g., Flowcept, W3C PROV) and data streaming tools (Kafka, Redis, RabbitMQ).
• Understanding of HPC workflow orchestration platforms such as Argo, CrewAI, Parsl, or RADICAL-Pilot.

Nice To Haves

• Knowledge of tools such as Grafana, Polars, or Pandas for monitoring and analyzing large-scale workflow execution and provenance data.
• Familiarity with synthetic workflows, graph-based reasoning, or computational chemistry/molecular dynamics workflows.
• Expertise in AI techniques such as retrieval-augmented generation (RAG), schema-driven reasoning, and graph traversal in provenance.
• Background in developing scalable tools for cross-domain, edge-to-HPC workflows using distributed architectures.
• Proven ability to integrate dynamic schema design and metadata enrichment into AI workflow systems.

Responsibilities

• Research and prototype LLM-driven agents capable of autonomous and interactive decision-making, anomaly detection, and guided experimentation in distributed scientific workflows.
• Design scalable systems for multi-workflow provenance capture, enhancing traceability, reproducibility, and transparency while facilitating intelligent multi-agent orchestration.
• Collaborate with domain scientists, computer scientists, engineers, and facility operators to integrate AI seamlessly into experimental and computational pipelines.
• Demonstrate the effectiveness of dynamic workflows in representative use cases such as materials discovery, combustion chemistry, and additive manufacturing.
• Publish research findings in peer-reviewed journals, conferences (e.g., SC, NeurIPS, AAAI), and open-source repositories.
• Mentor graduate students and contribute technical expertise to team projects aligned with ORNL’s strategic scientific goals.

What This Job Offers

• Job Type: Full-time
• Career Level: Entry Level
• Education Level: Ph.D. or professional degree
• Number of Employees: 5,001-10,000 employees