AgentStage: Exploiting LLM Thinking for Data Staging in Scientific Agents
Abstract
LLM agents on scientific data tasks typically spend most of their session wall time inside tool calls reading data from storage. Yet before each tool call, the agent reasons in a thinking phase, during which the storage layer sits idle and the next tool’s files begin to surface in the reasoning. Exploiting this is hard: the agent rarely names complete paths, so those files must be inferred and then staged before the tool fires. We present AgentStage, which reads the streaming reasoning and the agent’s filesystem probes to identify those files, then stages them into fast local storage with in-flight decompression so the tool’s reads land hot. We evaluate AgentStage on three curated I/O-bearing scientific tasks and one data-intensive task from each of three community benchmarks (MLE-bench, KramaBench, DSBench), across four reasoning models (three closed-weight, one open-weight). Under a paired trajectory-replay protocol that fixes each agent’s tool sequence and varies only cold versus staged execution, AgentStage delivers an average 1.74× end-to-end speedup on the curated tasks, up to 2.34× per run. On the three community benchmarks, AgentStage delivers up to 1.36× speedup. The results show that thinking-phase staging helps when I/O share is high and prefetch completes within the thinking window, while compute-bound or many-file workloads see smaller gains. We show both conditions can be estimated before any staging run, so AgentStage’s benefit is predictable in advance. AgentStage adds negligible LLM-side overhead and needs no per-model or per-task tuning.
Finding
Under a paired trajectory-replay protocol, staging the files that surface in an agent’s thinking phase delivers an average 1.74x end-to-end speedup on curated I/O-bearing scientific tasks, up to 2.34x per run, and up to 1.36x on three community benchmarks.
System
- 2024–present
IOWarp
The NSF-funded context-management program I proposed and architect across GRC, The HDF Group, and the University of Utah. IOWarp connects the HPC, storage, I/O, and data-systems lineage to scientific AI through the CLIO architecture.
More papers on IOWarp
Browse all papers on IOWarp- 2026
Trigger, Render, Reason: Multimodal LLM Agents for Science-Aware Scientific Simulation Monitoring
Hua Xu, Jaime Cernuda, Ana Gainaru, Scott Klasky, Anthony Kougkas, Xian-He Sun. International Conference for High Performance Computing, Networking, Storage, and Analysis (SC 2026)
- 2026
GPUH5: A GPU-first I/O Library for Scientific Data
J. Fernandes, Luke Logan, Xian-He Sun, Anthony Kougkas. 11th International Parallel Data Systems Workshop (PDSW 2026), held in conjunction with SC26
- 2026
Trigger-Gated Multimodal LLM Agents for In-situ Scientific Simulation Monitoring
Hua Xu, Jaime Cernuda, Anthony Kougkas, Ana Gainaru, Scott Klasky, Xian-He Sun et al. 11th International Parallel Data Systems Workshop (PDSW 2026), held in conjunction with SC26
- 2026
Type-Safe, Dynamically Defined Metadata Support for Scientific Applications
J. Teves, C. Brinker, Anthony Kougkas, Ana Gainaru, Scott Klasky, Xian-He Sun et al. 11th International Parallel Data Systems Workshop (PDSW 2026), held in conjunction with SC26
Citation
@inproceedings{yildirim2026agentstage,
author = {Yildirim, I. and Sun, X.-H. and Kougkas, A.},
title = {{AgentStage: Exploiting LLM Thinking for Data Staging in Scientific Agents}},
booktitle = {22nd IEEE International Conference on eScience (eScience 2026)},
year = {2026},
month = sep,
doi = {10.1109/eScience70237.2026.00040},
pages = {271--280},
url = {https://www.escience-conference.org/2026/papers},
publisher = {IEEE},
} Related notes
- 9 min
One destination, N tiers
Every tier supplies the application at once, and the application sees one destination. The thesis from a November 2026 talk to my group at the Gnosis Research Center, the six lines of I/O work behind it since 2016, and my bet that they are the foundation agents need for context.
- 4 min
SC26 in Chicago
What the group brings to SC26 in Chicago, the pressure from GPUs and agent fleets that runs through all of it, and an invitation to Thursday's Birds of a Feather on HDF5 in the age of AI-driven science.
- 6 min
CLIO as the entry point
CLIO should become the single place a developer starts with IOWarp, built as an application SDK with deployment profiles rather than an umbrella package or a gateway service. This is the direction for 2027; the API and the scope are still open.