LLM-agent researcher: autonomous experimentation for industrial recommenders and self-evolving agent environments
2
papers here
1
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-28) 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—
No output metrics are available in the current snapshot, so this axis is unscored. Once linked publication records accumulate, it would measure the volume and consistency of Fan Wu's papers, preprints, and other research artifacts over time.
Impact—
Citation-based impact cannot be computed from the reviewed-paper snapshot alone, so this axis is left unscored. It would measure citations, downstream uptake, and field influence of Fan Wu's work once linked citation records become available.
Focus100/100
The focus score of 100 reflects a field distribution containing a single recorded field — machine learning for recommender systems and LLM agents. Both cataloged papers stay inside one LLM-agent arc: autonomous experimentation for industrial recommender systems and environment engineering for recursive self-improvement. The topical lane is narrow and deliberate, trading breadth for depth in agentic machine-learning systems.
Collaboration69/100
A collaboration score of 69 rests on 31 distinct coauthors in the snapshot, with a recurring core of twelve frequent collaborators including Shuang Yang, Zijie Zhuang, and Changxin Lao. Both 2026 papers are team efforts in which Fan Wu holds middle or last authorship rather than first or corresponding, consistent with contributing inside larger project groups.
Activity—
Publication cadence and temporal throughput are not available in this snapshot, so the axis is unscored. It would measure how actively Fan Wu publishes across years; on record here are only two 2026 arXiv preprints.
Reproducibility20/100
Reproducibility scores 20, computed as a mean over a single assessed paper (reproPapers: 1), so the figure is fragile. In the catalog, the AgentX recommender paper is rated low readiness because private industrial data limits external verification, while Env-Rethink reaches medium. Wider artifact and data release would sharpen this axis.
Research summary
Fan Wu is a researcher working at the intersection of large language model agents and applied machine learning systems; this profile is built from two 2026 arXiv papers reviewed in the Evolve catalog. The recorded focus is unusually narrow: the field distribution contains a single recorded field — machine learning for recommender systems and LLM agents — and both papers extend one agentic arc. The first, 'Advancing Model Research in AgentX,' describes a long-horizon autonomy setup in which a dual-agent system completes 636 autonomous recommender experiments with candid internal accounting; the Evolve review notes that private industrial data limits external verification. The second, 'Breaking the Environment Wall,' introduces Env-Rethink, where a 27B verifier lifts noisy-environment agent pass rates by +13.3pp across 9 models while evolving harder tasks — a strong system whose recursive self-improvement claim the review treats as unproven. Collaboration is substantial: 31 distinct coauthors appear in the snapshot, with a recurring core of twelve frequent collaborators such as Shuang Yang, Zijie Zhuang, and Changxin Lao; on both papers Fan Wu is neither first nor corresponding author. Radar scores show maximal focus (100), solid collaboration (69), and low reproducibility (20, from one assessed paper), while output, impact, and activity are not yet scored for lack of data. No institutional affiliation is attached to this record.
Collaboration pattern
Fan Wu's recorded work is collaborative: 31 distinct coauthors appear in the snapshot, and a stable core of twelve frequent collaborators — including Shuang Yang, Zijie Zhuang, Changxin Lao, Pengbo Xu, and Hanwen Xu — recurs across projects. In both cataloged 2026 papers Fan Wu is neither first nor corresponding author (middle on AgentX, last on Env-Rethink), consistent with contributing key work inside larger engineering-driven teams rather than leading standalone efforts.
Research keywords
Highlights
Autonomous experimentation at industrial scale (AgentX)
Evolving agent environments (Env-Rethink)
Timeline
- 2026Co-authored 'Advancing Model Research in AgentX' on long-horizon autonomy for industrial recommender systems, in a middle author position; the dual-agent system ran 636 autonomous experiments.
- 2026Co-authored 'Breaking the Environment Wall' (Env-Rethink) as last author; a 27B verifier lifted noisy-environment agent pass rates by +13.3pp across 9 models.
- 2026Both papers entered the Evolve catalog with reproducibility readiness rated low and medium respectively, within a recorded collaboration network of 31 distinct coauthors.
Papers in the Evolve catalog
Frequent co-authors
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.
- Institution, country, and ORCID are not attached to this record; 'Fan Wu' is a common name, so the profile is name-resolved from public bibliographic data.
- Bibliometric scores are computed from public bibliometric aggregator data and cover only records linked to this name.
- Output, impact, and activity dimensions are unscored because citation counts and a full publication history were not available in this snapshot.
- The reproducibility mean (20) is computed from one assessed paper and may shift as more work is reviewed.