LLM multi-agent researcher building shared memory and human-AI collaboration for AI-driven research automation.
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-20) 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 computed for this profile: publication counts, venue distribution, and yearly output volume would measure research productivity, but the bibliographic records available here are insufficient to score it.
Impact—
Impact data is unavailable in this snapshot: citation counts, h-index, and downstream adoption would measure how widely the work influences others, but no citation records exist yet, so no score is given.
Focus0/100
Focus scores 0/100 in the radar, the lowest concentration band: the field distribution is an exact 50/50 split between 'AI agents and human-AI collaboration' and 'machine learning (LLM multi-agent systems, AI for research automation)'. With only two cataloged papers and a wide 69-coauthor network, the profile reads as broadly exploratory rather than concentrated; more publications would be needed to tell whether this is a deliberate dual-track agenda or early-stage drift.
Collaboration85/100
Collaboration scores 85/100, backed by 69 distinct coauthors — an unusually wide network for a two-paper profile. Twelve recurring collaborators are recorded, including Honglin Guo, Tao Gui, and Guanting Dong; the large, rotating teams on both 2026 papers fit the big-group style of LLM agent and model-release work, in which Dong held middle and last authorship rather than leading as corresponding author.
Activity—
Activity cannot be scored yet: yearly publication cadence and recent-output trends would measure how actively this researcher publishes, but the snapshot holds only two papers from a single year (2026), too little to establish a tempo.
Reproducibility38/100
Reproducibility scores 38/100, the mean across the two reviewed papers. 'Agora' rates medium — its Git-based provenance design speaks directly to reproducibility, though the causal benefit of the shared-memory approach over a plain log is untested — while 'Atria Dawn' rates low, with Evolve flagging cherry-picking risk in its benchmark table and self-reported attribution numbers. Artifact transparency is a visible theme, but not yet a demonstrated strength.
Research summary
Yi Dong works at the intersection of LLM multi-agent systems and AI-for-research automation, a profile built from two 2026 arXiv papers reviewed by Evolve plus a sparse bibliometric snapshot. The field distribution is an even 50/50 split between 'AI agents and human-AI collaboration' and 'machine learning (LLM multi-agent systems, AI for research automation),' indicating a program centered on agentic language models — how they share memory, accumulate verifiable experience, and collaborate with humans. On 'Agora: Git as Shared Memory for Collective AutoResearch' (2026), Dong is the last author: the paper frames a Git-committed research DAG as shared memory for LLM agents, and a 12-day, 13-agent run reaches 1.899 bpb, though Evolve notes the causal benefit over a plain log remains untested. On 'Atria Dawn: The Dawn of Agentic Superintelligence' (2026), a model release report, Dong is a middle author; Evolve calls its human-AI collaboration study the most interesting part while flagging cherry-picking risk in the benchmark table and self-reported attribution numbers. Collaboration scores 85 with 69 distinct coauthors; reproducibility scores 38 across the two papers; output, impact, and activity metrics are not yet available.
Collaboration pattern
Yi Dong's collaboration score is 85/100 with 69 distinct coauthors — a very wide network relative to the two-paper record. Twelve collaborators recur, including Honglin Guo, Tao Gui, Yicheng Chen, and Guanting Dong. The authorship pattern — middle on 'Atria Dawn', last on 'Agora', neither corresponding — suggests a contributor within large rotating LLM-agent teams rather than a project lead; the Evolve catalog records no first-author or corresponding-author paper for this name.
Research keywords
Highlights
Git as shared memory for a 13-agent auto-research pipeline
Human-AI collaboration study at the core of an agentic model release
Timeline
- 2026Middle author on 'Atria Dawn: The Dawn of Agentic Superintelligence', an agentic model release report covering a verifiable experience pipeline, recursive self-improvement, and a human-AI collaboration study.
- 2026Last author on 'Agora: Git as Shared Memory for Collective AutoResearch', which uses a Git-committed research DAG as shared memory for LLM agents; a 12-day, 13-agent run reaches 1.899 bpb.
- 2026Both cataloged papers land in the same year, drawing on a network of 69 distinct coauthors; the profile remains early-stage, with output, impact, and activity metrics not yet computed.
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.
- Bibliometric scores are computed from a sparse public bibliometric snapshot; only two reviewed papers exist in the Evolve catalog, so the output, impact, and activity axes are missing rather than low.
- 'Yi Dong' is a common name; this profile is name-resolved from public bibliographic data, and no institution, country, or ORCID was available to confirm identity.
- The reproducibility score (38) averages Evolve's readiness assessments of just two papers (one medium, one low); it is not a general reproducibility audit.
- Both papers are recent 2026 arXiv preprints with no peer-review or citation history in this snapshot.