Research Scientist, Agentic Workflows

Oak Ridge National Laboratory•Oak Ridge, TN

1d•Hybrid

About The Position

Oak Ridge National Laboratory (ORNL) is seeking a Research Scientist to focus on agentic workflow orchestration and cross-facility integration in support of the Genesis Mission, the American Science Cloud, and the Oak Ridge Leadership Computing Facility (OLCF). The role involves designing and deploying agentic workflows and frameworks that connect leadership computing resources, instruments, and cloud platforms. The scientist will collaborate with scientists, facility operators, and DOE laboratory partners to reduce barriers to scalable, automated scientific computing. The Workflows and Ecosystem Services (WES) Group, within which this role sits, specializes in designing and implementing workflow, data, and ecosystem technologies for reliable, automated execution of scientific applications across diverse computing environments, integrating various resources into cohesive systems. This group is part of the Advanced Technologies section of the National Center for Computational Sciences (NCCS) at ORNL, which provides advanced computational and data science infrastructure and hosts the OLCF, home to the exascale supercomputer Frontier.

Requirements

PhD. in Computer Science, Computer Engineering, or a closely related field.
At least 3 years of experience designing or operating distributed systems, provenance frameworks, or workflow orchestration frameworks.
At least 2 years of experience designing agentic workflows.
At least 2 years of programming experience in Python, with working familiarity in Bash and/or a compiled language (C/C++ or similar).
Experience with Linux-based HPC environments and familiarity with job scheduling systems (e.g., SLURM or PBS).
Strong verbal and written communication skills, with the ability to collaborate across technical and scientific teams.

Nice To Haves

Demonstrated experience with provenance models such as W3C PROV.
Demonstrated experience publishing research in high-impact venues such as SC, IPDPS, HPDC, or IEEE eScience.
Experience working in or coordinating across DOE national laboratory environments, particularly in the context of integrated or cross-facility research infrastructure.
Background in scientific data management, provenance tracking, or metadata systems.
Demonstrated ability to lead technical projects, manage milestones, and coordinate multidisciplinary teams.
Familiarity with basic cyber-security principles (e.g., SSH key hygiene, least privilege, network segmentation) as applied to workflow and API design.

Responsibilities

Lead the research and development of runtime data integration and provenance capture systems (e.g., Flowcept) for distributed scientific workflows, ML training pipelines, and agentic systems.
Design low-overhead instrumentation and adapters for common execution frameworks, define provenance data models aligned with community standards (e.g., W3C PROV), and build query, storage, and streaming layers for workflow telemetry.
Drive the technical roadmap, release process, and architecture of the software.
Serve as ORNL's technical lead for cross-facility initiatives (e.g., AmSC, Genesis Mission) building interoperable provenance and workflow-data capabilities across leadership computing, experimental facilities, and cloud platforms.
Design services to track data lineage, model artifacts, and computational campaigns across the DOE complex, coordinating with partner laboratories to align schemas, APIs, and metadata standards.
Define technical scope, coordinate multi-institutional development teams, track milestones and deliverables, and represent ORNL in project-level technical planning.
Contribute to proposals and reports to sustain and grow the portfolio.
Serve as solutions architect for OLCF platform efforts (e.g., OPAL), translating requirements into deployable system designs.
Evaluate architectural trade-offs across orchestration, data management, storage, and security; produce reference architectures and integration patterns; and work with facility operations staff to move designs from prototype to production.
Design and implement agentic workflow architectures for autonomous, closed-loop scientific experimentation across HPC and instrument facilities.
Integrate agentic frameworks with provenance and orchestration layers for AI agents to plan, execute, monitor, and adapt multi-step workflows while producing auditable records.
Apply these patterns to use cases such as autonomous materials discovery, self-steering simulations, and adaptive experimental campaigns.
Own the technical direction and health of open-source research software, including code review, CI/CD, testing, packaging, documentation, and release management.
Grow and support an external contributor and user community, respond to issues, and ensure software deployability beyond ORNL.
Establish and sustain research collaborations with universities, national laboratories, and international partners.
Define joint research directions, co-supervise students, interns, and postdocs, and convert collaborative work into shared software and publications.
Collaborate with domain scientists in areas like materials science and climate science to translate research objectives into scalable, automated workflow and data-capture solutions.
Provide documentation, training, and direct user support.
Conduct original research on provenance, workflow systems, orchestration, and data management for AI-driven science.
Publish in high-impact peer-reviewed venues and present at major conferences.