AI evaluation researcher at A*STAR: LLM multi-agent trading safety and medical vision-language assessment
2
papers here
2
reviewed by Evolve
0
first-author
1
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) 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—
The output axis is not yet scored: publication counts, venue distribution and yearly productivity data are missing from the current snapshot. Once a fuller bibliographic record is linked, this axis will measure the researcher's total publication volume and production rate.
Impact—
Impact is not yet scored because citation counts, h-index and venue-level influence are absent from the available data. This axis would measure how widely the two reviewed papers and the broader record are cited and taken up by subsequent work.
Focus0/100
Focus currently scores 0 because the reviewed output is split evenly across two distinct domains: 50% on AI safety for LLM multi-agent trading systems together with quantitative finance, and 50% on medical image analysis with AI evaluation methodology. This even split from the field distribution signals an exploratory, dual-track portfolio rather than a concentrated agenda.
Collaboration41/100
Collaboration scores 41, based on 7 distinct coauthors across the two reviewed papers. The coauthor pool is small and includes recurring partners such as Cheston Tan, Qi Rong Sua, Junhao Dong, Nguyen Duc Thai, Yuqing Wen, Flint Xiaofeng Fan and Roger Wattenhofer, pointing to compact working teams rather than large consortia.
Activity—
Activity is not scored yet: the current snapshot lacks a full yearly publication record, so recent output cadence cannot be computed. This axis would measure publishing tempo and continuity over time once more of the bibliographic record is linked.
Reproducibility78/100
Reproducibility scores 78, the mean readiness across 2 reviewed papers. The medical VLM evaluation paper (AuthEval) rates high, while the trading-contagion paper rates medium, held back by a single ticker-month setup and missing variance data. Overall this indicates above-average attention to evaluable, checkable methods.
Research summary
Yew-Soon Ong is a researcher at the Agency for Science, Technology and Research (A*STAR) whose cataloged work to date spans two distinct domains in equal measure: AI safety for LLM-based multi-agent trading systems, and evaluation methodology for medical vision-language models. The Evolve catalog holds two 2026 arXiv papers. In the first, as last and corresponding author, Ong's team studies how poisoned social-media signals spread through multi-agent trading systems, proposing contagion metrics; Evolve judges the framework solid while flagging the single ticker-month scope and absent variance data. The second paper, Authority-Preserving Evaluation of Medical Vision-Language Assistants, introduces AuthEval, which formalizes the gap between what a model proposes and what a human selects in teledermatology triage; it is rated sound and honest with high reproducibility readiness, though its evidence is entirely simulated. Bibliometrically, the profile is compact: 7 distinct coauthors across the two papers, a collaboration score of 41, a reproducibility score of 78, and a focus score of 0 reflecting the even 50/50 field split. Recurring coauthors include Cheston Tan, Qi Rong Sua, Junhao Dong, Nguyen Duc Thai, Yuqing Wen, Flint Xiaofeng Fan and Roger Wattenhofer. Output, impact and activity axes lack sufficient data and are not scored. This profile is name-resolved from public bibliographic records linked to ORCID 0000-0002-4480-169X and reflects only the reviewed slice of work.
Collaboration pattern
Collaboration runs in small, stable teams: the two reviewed 2026 papers share a pool of 7 distinct coauthors, yielding a collaboration score of 41. Recurring partners include Cheston Tan, Qi Rong Sua, Junhao Dong, Nguyen Duc Thai, Yuqing Wen, Flint Xiaofeng Fan and Roger Wattenhofer, spanning both the trading-safety and medical-evaluation projects. The moderate score reflects this compact circle rather than broad cross-institution networks.
Research keywords
Highlights
First systematic look at poisoned social-media signals in LLM multi-agent trading
AuthEval: formalizing the authority gap in medical VLM triage
Timeline
- 2026Last and corresponding author on 'Contagion on the Trading Floor', proposing a framework for how adversarial social-media signals spread in LLM multi-agent trading systems, with contagion metrics.
- 2026Co-authored 'Authority-Preserving Evaluation of Medical Vision-Language Assistants' as a middle author, contributing to the AuthEval methodology for medical VLM triage.
- 2026Both papers entered the Evolve catalog, covering two distinct fields at a 50/50 split with a mean reproducibility readiness of 78 across 2 papers.
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.
- Radar scores come from Evolve's OpenAlex-derived bibliometrics; the output, impact and activity axes lack sufficient underlying data and are unscored.
- Only two arXiv preprints from 2026 have been reviewed; they may not represent the full publication record and are not necessarily peer-reviewed.
- The profile is name-resolved from public bibliographic records linked to ORCID 0000-0002-4480-169X; coverage may be incomplete where records are not linked to that identity.
- No citation counts, h-index, venue history or career details were available in the input, so none are quoted.