Agents powered by AI/ML are becoming ingrained in orchestration. Dispatch of work is the task of receiving a request, transforming it for a workload manager, and successfully submitting it. Running scientific workflows across multi-cluster environments introduces substantial challenges of dynamic job transformation, dispatch, and submission to heterogeneous clusters. These tasks are well-suited to agents, which can receive textual instructions for work, prepare job specifications, and dispatch. In this work, we assess the reliability of a dispatch agent across 432 runs, testing all possible combinations of five feature dimensions across four prompt styles. The agent is highly reliable (97.9% success). We test a full orchestration to submit, queue, match, score, select, transform, and dispatch in a multi-cluster experiment. We find that descriptive metadata increases successful execution from 48% to 87% of 220 submitted jobs, eliminating architecture mismatch, and improving performance for five of ten measurable applications by up to 3.3x.
High Performance Computing (HPC) centers are expanding to encompass resources that extend beyond traditional computing. By extending resources to quantum computing, hybrid quantum-classical workflows tackle complex optimization problems that have never before been possible. However, integrating quantum processing units (QPUs) into cloud-native and scientific workload managers presents a unique orchestration challenge: remote quantum devices introduce a second, external queue -- a two-queue problem -- alongside the queue owned by the traditional scheduler. In this work we present Fluence, a Kubernetes scheduler plugin backed by the Fluxion graph-based scheduler, that enables informed, gang-scheduled placement for quantum-classical workloads and custom resources. We evaluate Fluence across three scenarios using AWS Braket simulators and real QPUs. First, under node contention, Fluence's atomic gang placement all but eliminates the wasted node-time that a default scheduler accrues by partially placing gangs. Second, we introduce a synchronization primitive for the two-queue problem in which a single producer submits a shared quantum task while consumers remain scheduling-gated, reducing worker idle time by roughly 5x under short device queues and by orders of magnitude when a real device queue stretched to hours. Third, cost- and queue-aware backend selection pins the cheapest or shortest-queue device satisfying a workload, cutting mean per-run cost by roughly 70x and time-to-result from hours to under a minute. Together, these results show that quantum-awareness can be added to a cloud-native scheduler without modifying user containers.
The means to execute and orchestrate software components has changed from human-written code to descriptive prose. In high performance computing, this transition is represented in application orchestration, workload management, and system monitoring and debugging, to name a few. The underlying means to enable descriptive definition of tasks is the use of the Large Language Model with associated tool functions and resources. A combination of a model with access to such resources, modeled in software, encompasses an autonomous framework. As fully automated and agentic frameworks are developed for science, it is important to assess reliability and strategies scoped to specific tasks. In this work, we assess the extent to which an agentic framework can optimize and run an HPC scaling study with a low latency network in Amazon Web Services, accurately transform HPC job specifications between workload managers, and design and run an entire biosciences workflow. We find that the framework completes all three tasks while surfacing task-specific failure modes. In the scaling study, agents deploy and optimize applications but monitor running jobs inefficiently, preferring conservative fixed waits over event subscriptions. In job translation, they convert specifications between Slurm and Flux with high accuracy, with processor-affinity flags the most common error. In the bioscience workflow, the agent reproduces an expert-written variant-calling pipeline almost exactly -- agreeing with the reference call set in 18 of 19 completed runs -- and reaches this result through many distinct yet functionally equivalent workflow implementations. This information is invaluable moving forward to developing multi-cluster setups with scheduling and transformation handled by agents.
Large Language Models (LLMs) are serving as a catalyst of change for research practices, touching the daily lives of staff scientists, software engineers, and system administrators. The developments promise new degrees of autonomy, where categories of human work and decision making are replaced by autonomous, goal-oriented systems. This transition necessitates novel architectural paradigms and solid understanding of the strengths and limitations of LLMs. In this work, we design agents to intelligently deliver the entire life-cycle of an HPC application experimental run in cloud -- creation and build of a container, deployment in Kubernetes, optimization, and orchestration of a scaling study. We pursue this task for four well-known HPC applications to build multi-platform images and optimize across 21 instance types in Kubernetes. We demonstrate successful linear scaling with patterns approved by human experts, designs that improve work time to completion, and review suggested best practices for agentic design and collaboration.
This paper presents a hierarchical, dynamic architecture and software to discover resources across diverse cloud, edge, and HPC systems and exemplifies the importance of careful coordination between agents, discovery tools, and infrastructure for agentic science.