Early-career Westlake University researcher spanning ML systems and scientific machine learning
2
papers here
2
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
0
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-09) 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 scoring is unavailable: the linked OpenAlex record shows h-index 0, i10-index 0, and yearly counts for 2026 only (3 works), far too sparse to summarize. This axis would measure cumulative publication volume, venues, and authorship roles.
Impact—
Impact cannot be scored yet: with h-index 0, i10-index 0 and no citation history in the record, there is no citation signal to aggregate. This axis would track citations, uptake, and influence of the 2026 preprints as they accrue.
Focus0/100
The focus score is 0, reflecting an evenly split portfolio: the OpenAlex field distribution assigns 50% to ML systems (inference serving / cloud computing) and 50% to machine learning for physical systems, and five topics appear only once or twice each. The two 2026 papers confirm the dual track — a first-authored LoRA serving trace study and a middle-authored Bayesian filtering paper in separate fields.
Collaboration39/100
Collaboration scores 39 on this record: 6 distinct coauthors appear alongside 3 works in the recent three-year window. Recurring partners include Bin Liao, Tao Zhou, Yanping Liu, Ruiqi Feng, Chongyi Wang, and Tailin Wu. Authorship spans both roles — first author on the LoRA serving study and middle author on the flow-matching filtering paper — without a corresponding-author credit yet.
Activity40/100
Activity scores 40: OpenAlex lists 3 works in the recent three-year window, with per-year counts covering only 2026 (3 works). Both catalogued papers appeared in 2026 on arXiv, indicating a concentrated current burst rather than a long sustained record; h-index 0 and i10-index 0 confirm an early-stage profile.
Reproducibility78/100
Reproducibility is the strongest axis at 78, averaged over 2 reviewed papers. The first-authored LoRA co-occurrence trace study is rated high readiness — a transparent trace analysis, though limited to 1,090 multi-adapter requests — while the flow-matching Bayesian filtering paper is rated medium, with Evolve noting thin particle baselines and error bars.
Research summary
Tao Zhang is a researcher at Westlake University (China) whose profile is name-resolved from public bibliographic data — Tao Zhang is a common name, and no ORCID is linked. The record is early-stage: OpenAlex lists an h-index of 0, an i10-index of 0, and per-year publication counts covering only 2026 (3 works), with 3 works in the recent three-year window. Two 2026 arXiv papers reviewed by Evolve define the visible research span, and it is evenly split: 50% ML systems (inference serving / cloud computing) and 50% machine learning for physical systems. The first-authored paper is a trace study of LoRA adapter co-occurrence in production diffusion-model inference services, finding usage sparse, model-driven, and preloadable; Evolve rates its reproducibility high while noting the analysis covers 1,090 multi-adapter requests. The second paper, in middle authorship, introduces BFF, which encodes Bayesian filtering posteriors in test-time trained flow-matching weights and reports strong PDE and tokamak results, with Evolve flagging thin particle baselines and error bars. Reproducibility is the strongest radar dimension at 78 (mean over 2 reviewed papers), while collaboration (39) and activity (40) are modest, and output and impact cannot be scored from current data. Frequent coauthors include Bin Liao, Tao Zhou, Yanping Liu, Ruiqi Feng, Chongyi Wang, and Tailin Wu.
Collaboration pattern
The record shows 6 distinct coauthors across 3 recent works, with a collaboration score of 39. Recurring partners include Bin Liao, Tao Zhou, Yanping Liu, Ruiqi Feng, Chongyi Wang, and Tailin Wu. Authorship spans both ends: first author on the LoRA serving trace study and middle author on the flow-matching filtering paper, with no corresponding-author credit in the catalog yet.
Research keywords
Highlights
LoRA adapter co-occurrence in production diffusion serving
BFF: test-time trained flow matching for Bayesian filtering
Evenly split 2026 output across two fields
The OpenAlex record logs 3 works in 2026 — the only year with per-year counts — and the field distribution splits 50/50 between ML systems (inference serving / cloud computing) and machine learning for physical systems.
Timeline
- 2026Posts first-authored trace study of LoRA adapter co-occurrence in production diffusion inference services to arXiv.
- 2026Co-authors (middle position) the BFF paper on Bayesian filtering via test-time trained flow matching for physical systems.
- 2026OpenAlex logs 3 works in 2026, the only year with per-year counts, with 6 distinct coauthors on record.
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.
- Bibliometrics come from OpenAlex only; no ORCID is linked, and Tao Zhang is a common name, so identity resolution should be treated with caution.
- Career-long metrics are effectively absent: h-index and i10-index are 0, and per-year counts cover 2026 only.
- Only 2 papers have been reviewed by Evolve; radar scores rest on this small sample.
- Both catalog papers are arXiv preprints; peer-review status and final venues are not reflected.