First-author builder at the interface of LLM evaluation auditing and LLM-augmented Bayesian network structure learning
2
papers here
2
reviewed by Evolve
2
first-author
0
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—
Not yet scored: the snapshot contains no publication-count or venue data beyond the two arXiv preprints reviewed here, so total scholarly output cannot be computed. This axis would measure the volume and consistency of papers over time.
Impact—
Not yet scored: no citation counts, h-index, or adoption signals are available for this profile. This axis would measure citations, downloads, and uptake of the reviewed work by other researchers and systems.
Focus0/100
The radar records focus at 0, driven by the raw field distribution: research effort splits exactly 50/50 between natural language processing (LLM-as-a-judge evaluation) and machine learning (probabilistic graphical models and LLM-augmented causal discovery). With only two papers, this even split reads as maximal breadth and minimal concentration — MAWILE audits LLM judges while ABSOL embeds LLMs inside Bayesian network structure learning. More publications would be needed to tell whether this is a deliberate dual-track agenda or early-stage exploration.
Collaboration32/100
The radar scores collaboration at 32, based on 4 distinct coauthors across the reviewed work. The recurring collaborators are Estevam Hruschka, Farima Fatahi Bayat, Pouya Pezeshkpour, and Chen Shen, and both catalog papers are first-authored without a corresponding-author flag, suggesting a hands-on contributor role within a small, stable team rather than a broad network. The low distinct-coauthor count keeps this score modest; it reflects a compact collaboration footprint, not a weakness of the work itself.
Activity—
Not yet scored: only two 2026 preprints are visible, too few data points to gauge publishing cadence. This axis would measure recent publication rate and sustained research engagement over time.
Reproducibility78/100
Reproducibility is the strongest scored axis, at 78 — the mean readiness rating across 2 reviewed papers. ABSOL is rated high readiness (LLM-guided Bayesian subsampling evaluated on synthetic benchmarks), while MAWILE is rated medium, consistent with the review note that its evidence is broad but statistically thin. Together they suggest code- and benchmark-oriented work habits; the rating rests on only two papers, so it should be read as promising rather than settled.
Research summary
Jackson Hassell is an early-career researcher whose public bibliographic footprint, anchored to ORCID 0009-0004-5573-2759, currently consists of two first-author arXiv preprints, both posted in 2026. His work occupies a distinctive junction: large language models are treated both as objects of audit and as bounded algorithmic components inside rigorous pipelines. The first strand, MAWILE (2026), is a workbench that unifies judge-audit perturbations across prompt, rubric, input, and output axes for LLM-as-a-judge evaluation; Evolve's review calls it a well-engineered system paper with broad but statistically thin evidence. The second strand, ABSOL (2026), uses aggregated Bayesian subsampling orchestrated with LLM guidance to scale Bayesian network structure learning to 1041 nodes; gains on synthetic benchmarks look real, though the largest-scale runs lack variance estimates. The radar records a reproducibility mean of 78 across the two reviewed papers (ABSOL high, MAWILE medium), a collaboration score of 32 built on four distinct coauthors, and an even 50/50 field split between natural language processing and machine learning with probabilistic graphical models. No citation, venue, or affiliation data is available in this snapshot, so output, impact, and activity remain unscored.
Collaboration pattern
Jackson Hassell's visible collaboration footprint is compact: 4 distinct coauthors, with Estevam Hruschka, Farima Fatahi Bayat, Pouya Pezeshkpour, and Chen Shen recurring across the catalog. Both papers are first-authored without a corresponding-author flag, consistent with a hands-on contributor working inside a small, stable team — likely spanning both the NLP-evaluation and probabilistic-graphical-model strands — rather than a large distributed consortium.
Research keywords
Highlights
MAWILE: one workbench auditing LLM judges across four axes
ABSOL scales Bayesian network structure learning to 1041 nodes with bounded LLM guidance
A common thread: LLMs embedded in, and audited by, rigorous pipelines
Across both 2026 first-author papers, LLMs appear either as the object of meta-evaluation (MAWILE's judge audits) or as a bounded component inside classical structure-learning machinery (GES with BDeu scoring in ABSOL) — a dual perspective on where LLMs can and cannot be trusted.
Timeline
- 2026First-authored MAWILE preprint posted on arXiv, unifying perturbation-based audits of LLM-as-a-judge evaluators across prompt, rubric, input, and output axes.
- 2026First-authored ABSOL preprint posted on arXiv, scaling Bayesian network structure learning to 1041 nodes via LLM-orchestrated aggregated subsampling.
- 2026Both catalog papers land in the same year across two distinct fields, with 4 distinct coauthors and a mean reproducibility readiness of 78 — a concentrated, dual-track debut year.
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 are drawn from OpenAlex-based public bibliographic data; only the two reviewed arXiv preprints are in scope, so citation, venue, and h-index signals are absent.
- Profile is name-resolved via ORCID 0009-0004-5573-2759; no institutional affiliation or country is available in the data.
- Radar scores rest on very few papers (reproducibility from 2 papers, focus from a two-field split), so scores may shift substantially as the record grows.
- Reproducibility ratings and verdicts reflect Evolve's own paper reviews, not independent replication attempts.