Chen Dinachi

—

Unclaimed · AI-structured profileOpenAlex

Chen Dinachi — synthetic enterprise-data benchmarking with exact ground truth for LLM agent evaluation

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-16) 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 output axis would measure publication volume and productivity over time. Only three 2026 arXiv preprints are on record, too small a base for a stable output estimate.

Impact—

Not yet scored: the impact axis would measure citation-based influence such as total citations, h-index, and field-weighted attention. The recorded h-index and i10-index are both 0, so no meaningful impact signal exists yet.

Focus0/100

Focus scores 0 because OpenAlex splits the three works evenly across three field labels — LLM agent benchmarking, LLM agent evaluation, and synthetic data generation for agentic AI, each at one third. Read substantively, the spread is nominal: all three papers serve one tightly coupled program around the Era by Eon benchmark family, with supporting topics in topic modeling, AI applications, and NLP techniques.

Collaboration39/100

Collaboration scores 39. OpenAlex records 6 distinct co-authors, and the same six-person team — Benjamin Gruenbaum, Doron Porat, Assaf Natanzon, Roy Zavida, Or Itzahary, Omer Niv — recurs across all three papers. Dinachi sits in a middle author position on each and is not the corresponding author on any, indicating a stable in-group contributor role within a small, closed team rather than an expanding external network.

Activity40/100

Activity scores 40. The OpenAlex counts-by-year record shows all 3 works appearing in 2026, with 3 works in the last three years — the entire visible output lands in a single year, consistent with a research program that has just become publicly visible. No earlier publication record exists in the snapshot to indicate sustained multi-year activity, so the score remains modest.

Reproducibility43/100

Reproducibility averages 43 across the 3 reviewed papers. Evolve's verdicts pinpoint concrete gaps: the hidden-knowledge extension uses a single-seed evaluation and contains internal count errors; the synthesis paper's reference-free evaluation is circular and lacks external validation; the core benchmark paper shows sound validation but evaluates one small estate with code unreleased. Reproducibility readiness is rated medium, medium, and low respectively.

Research summary

Chen Dinachi is an early-stage researcher working at the intersection of LLM agent evaluation and synthetic enterprise data. The Evolve catalog records three 2026 arXiv preprints, all co-authored within the same six-person team — Benjamin Gruenbaum, Doron Porat, Assaf Natanzon, Roy Zavida, Or Itzahary, and Omer Niv — with Dinachi in a middle author position on each. The work centers on the Era by Eon benchmark family: a seed-generated fictional enterprise estate with exact ground-truth answers for benchmarking LLM agents ('The Era by Eon Benchmark'); a deterministic multi-system business-data synthesis pipeline with reference-free evaluation ('Generating a Consistent Enterprise'); and an extension adding eight deterministic hidden-knowledge questions to the benchmark ('Era by Eon: Benchmarking Enterprise Agents on Hidden Knowledge'). Field attribution splits evenly — one third each across LLM agent benchmarking, LLM agent evaluation, and synthetic data generation for agentic AI — reflecting one tightly coupled research program rather than scattered interests. OpenAlex currently records 3 works in 2026, 6 distinct co-authors, and an h-index of 0, so citation-based impact cannot yet be assessed. Evolve's reproducibility reviews average 43 across the three papers, flagging single-seed evaluation, circular reference-free evaluation, and unreleased code. This is an early-stage profile built from a thin bibliographic footprint; no institutional affiliation is on record.

Collaboration pattern

OpenAlex attributes 6 distinct co-authors, and the pattern is strikingly closed: the same six-person team — Benjamin Gruenbaum, Doron Porat, Assaf Natanzon, Roy Zavida, Or Itzahary, Omer Niv — appears on all three 2026 papers. Dinachi holds a middle author position on each and is corresponding author on none, consistent with a fixed internal team producing a single benchmark product line rather than outward-facing collaboration.

Research keywords

LLM agentsAgent benchmarkingSynthetic enterprise dataSynthetic data generationExact ground truthBenchmark generationReference-free evaluationHidden knowledgeTool use evaluationDeterministic generationEntity graphQuestion-conditioned generation

Highlights

Exact Ground Truth for Agent Benchmarks via a Generated Enterprise Estate
Seed-generates fictional companies with exact answer keys so LLM agents can be scored deterministically on systems-of-record questions; Evolve finds the validation sound but notes the headline evaluation covers one small estate and the code is unreleased.
Deterministic Synthesis and Reference-Free Evaluation of Multi-System Business Data
Builds a deterministic synthetic enterprise with an entity graph, adversarial detection, and question-conditioned generation, paired with reference-free evaluation; Evolve calls the engineering sound but flags that the evaluation is circular and lacks external validation.
Hidden-Knowledge Questions Added to the Era by Eon Agent Benchmark
Extends the benchmark with eight deterministic hidden-knowledge questions targeting enterprise data; Evolve praises the construct's elegance and honest limits, while noting that single-seed evaluation and internal count errors weaken the claims.

Timeline

2026First visible arXiv work: The Era by Eon Benchmark introduces a seed-generated fictional enterprise estate with exact ground truth for benchmarking LLM agents.
2026Generating a Consistent Enterprise presents deterministic multi-system business-data synthesis with an entity graph and reference-free evaluation.
2026Era by Eon: Benchmarking Enterprise Agents on Hidden Knowledge adds eight deterministic hidden-knowledge questions to the benchmark.
2026OpenAlex records 3 works this year with 6 distinct co-authors; the h-index stands at 0, so impact is not yet measurable.

Papers in the Evolve catalog

Frequent co-authors

Benjamin Gruenbaum×3 Doron Porat×3 Assaf Natanzon×3 Roy Zavida×3 Or Itzahary×3 Omer Niv×1

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.

All bibliometrics are derived from OpenAlex only; no ORCID or institutional affiliation is available in the source record.
The three cataloged works are 2026 arXiv preprints; citation-based impact has not accumulated (h-index and i10-index are 0).
This profile is name-resolved from public bibliographic data; the name may be shared by more than one person.
The reproducibility assessment covers only the three papers Evolve has reviewed.