Ian Foster

Argonne National Laboratory · United States

Unclaimed · AI-structured profileOpenAlex

Argonne researcher bridging HPC and AI infrastructure via agentic resource control and foundation-model virtualization.

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-16) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.

Research profile

Output—

Output metrics are not yet available: the snapshot's h-index and i10-index fields read 0, indicating the citation record was not resolved for this author name. This axis would measure total publications and per-year production once identity linkage is confirmed.

Impact—

Impact data is insufficient: citation counts, h-index, and i10-index are absent or zero in the current snapshot, so no influence score is computed. This axis would measure citation accumulation, citation velocity, and uptake of the reviewed papers.

Focus0/100

The focus score is 0, and the raw field distribution explains why: work splits exactly evenly — 0.5 to high-performance computing (agentic systems / distributed systems) and 0.5 to machine learning systems (MLSys) / AI infrastructure. Topic clusters likewise spread across five areas, led by scientific computing and data management (8), cloud computing and resource management (7), and distributed and parallel computing systems (7). The profile reads as deliberately broad rather than narrowly specialized.

Collaboration61/100

Collaboration scores 61 on the strength of 20 distinct coauthors in the recent window. The recurring partner list includes Kyle Chard, Yadu Babuji, Volodymyr Kindratenko, Shivaram Venkataraman, and Zhao Zhang, alongside newer co-authors such as Haotian Xie, Junlin Chen, and Mingkai Zheng — a team-scale, systems-building working pattern rather than solitary authorship.

Activity86/100

Activity is the strongest axis at 86. The snapshot records 18 works in the last three years, with countsByYear showing 1 work in 2025 jumping to 17 in 2026 — a steep, concentrated ramp. Both Evolve-reviewed papers date from 2026, consistent with an aggressive current publication pace in preprint form.

Reproducibility38/100

Reproducibility averages 38 across the two papers Evolve has reviewed (reproPapers = 2, reproMean = 38). The split is instructive: the foundation-model operating-system position paper is rated low readiness, offering diagnosis and agenda without prototype or evaluation, while Diamond Agent is rated medium, with live four-site results that remain small-scale.

Research summary

This profile is name-resolved from OpenAlex bibliographic data for Ian Foster at Argonne National Laboratory. The observable record sits at the junction of high-performance computing and machine-learning infrastructure: the field distribution splits evenly (0.5/0.5) between HPC (agentic systems / distributed systems) and MLSys / AI infrastructure, and the largest topic clusters are scientific computing and data management (8 works), cloud computing and resource management (7), and distributed and parallel computing systems (7). Tracked output is recent and accelerating — 18 works in the last three years, rising from 1 in 2025 to 17 in 2026 — which supports an activity score of 86, while 20 distinct coauthors yield a collaboration score of 61. Evolve has reviewed two 2026 papers: a position paper arguing for a self-evolving operating-system layer that virtualizes foundation models (judged compelling in its fragmentation diagnosis but lacking any prototype, evaluation, or evidence), and Diamond Agent, which demonstrates one-agent control of federated HPC across four live sites using typed skills and event-driven continuation (encouraging, but small-scale). The focus score of 0 reflects this deliberately broad two-field spread rather than narrow specialization, and citation-based output and impact metrics remain unavailable in the current snapshot.

Collaboration pattern

Collaboration is team-scale: 20 distinct coauthors in the recent window support the 61 score. A recurring core spans Kyle Chard, Yadu Babuji, Zhao Zhang, Volodymyr Kindratenko, Shivaram Venkataraman, and Max Burnette, while the 2026 papers bring a newer cluster including Haotian Xie, Junlin Chen, Mingkai Zheng, Yifan Zhu, Minu Mathew, and Suparna Bhattacharya.

Research keywords

foundation modelsagentic AIoperating system abstractionvirtualizationcompound AI systemscontext and memory managementhigh-performance computingfederated resource managementmulti-site job placementevent-driven agentsGlobus Computescientific computing

Highlights

A case for an operating-system layer to virtualize foundation models
A 2026 position paper proposes an FMOS: a self-evolving OS-style layer for foundation models, agentic AI, and compound AI systems, with context and memory management as core services. Evolve finds the fragmentation diagnosis compelling but notes there is no prototype, evaluation, or evidence yet; reproducibility readiness is rated low.
Single-agent control of federated HPC across four live sites
Diamond Agent (2026) shows one agent operating multiple supercomputers through typed skills and event-driven continuation, handling federated resource management and multi-site job placement atop Globus Compute. Live four-site results are described as encouraging but small-scale; reproducibility readiness is rated medium.
A sharp 2026 publication ramp
The tracked window shows output concentrated almost entirely in one year: countsByYear records 1 work in 2025 and 17 in 2026, within 18 works over the last three years. Both reviewed papers fall in this 2026 surge, marking an agenda-setting phase rather than an established citation record.

Timeline

2025First tracked work appears in the OpenAlex window (1 work), marking the start of the observable recent activity for this name-resolved profile.
2026Output accelerates sharply to 17 works in a single year, the peak of the tracked window and the basis of the 86 activity score.
2026The position paper argues for virtualizing foundation models with a self-evolving operating-system layer; Evolve finds the diagnosis compelling but no prototype or evaluation, rating reproducibility readiness low.
2026Diamond Agent demonstrates one-agent, multi-supercomputer control across four federated HPC sites via typed skills and event-driven continuation; readiness is rated medium.

Papers in the Evolve catalog

Frequent co-authors

Haotian Xie×1 Junlin Chen×1 Mingkai Zheng×1 Yifan Zhu×1 Minu Mathew×1 Max Burnette×1 Yadu Babuji×1 Volodymyr Kindratenko×1 Shivaram Venkataraman×1 Kyle Chard×1 Zhao Zhang×1 Suparna Bhattacharya×1

About this profile

This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.

Bibliometrics are computed from an OpenAlex snapshot only; no ORCID link was available, and h-index/i10-index read 0, so output and impact axes are marked missing.
The profile is name-resolved from public bibliographic data; the name is shared by multiple researchers, so some indexed works may belong to different individuals.
Counts reflect only the tracked recent window (18 works; countsByYear for 2025-2026), so longer-term productivity is not visible in this data.
Reproducibility scores come from Evolve's review of only 2 papers, both 2026 arXiv preprints.