Johnny Greco

—

Unclaimed · AI-structured profileOpenAlex

ML systems researcher; first author of NeMo Data Designer for multimodal synthetic data generation

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-09-21) 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: only one reviewed paper is on record, too small a base to measure publication volume, rate, or venue mix. It would quantify productivity from paper counts and authorship roles.

Impact—

Impact is not yet scored: citation counts, citation velocity, and downstream usage are unavailable in the current snapshot. This axis would measure how far the NeMo Data Designer work and other outputs propagate through the community.

Focus100/100

Scored 100: every cataloged paper sits in one field, machine learning systems (synthetic data generation for LLMs), giving a perfectly concentrated field distribution. With one reviewed paper the signal is real but early, marking a sharply defined specialization rather than a long-term pattern.

Collaboration48/100

Scored 48 from bibliometric data: 10 distinct coauthors appear in the record, and the frequent-coauthor list is dominated by a tight core — Nabin Mulepati, Andre Manoel, Eric Tramel, and Yev Meyer among them — suggesting concentrated team-based work rather than broad cross-institutional networks.

Activity—

Activity is not yet scored: the record spans too short a window to assess publishing cadence, recent engagement, or year-over-year output. It would track how consistently this researcher publishes and stays active over time.

Reproducibility55/100

Scored 55 as the mean over one reviewed paper. NeMo Data Designer carries a medium reproducibility readiness: Evolve credits an open-source, well-engineered framework with real case studies, but flags that reported gains lack ablations and uncertainty quantification — the main gap to turnkey replication.

Research summary

Johnny Greco is a machine learning systems researcher whose reviewed work centers on synthetic data generation for large language models. The Evolve catalog records him as first author of NeMo Data Designer: An Extensible Framework for Multimodal Synthetic Data Generation (2026, arXiv), an open-source framework that generates multimodal LLM training data through declarative pipelines and connects to the Nemotron ecosystem. Evolve's review calls the framework well-engineered and backed by real case studies, while noting that reported gains lack ablations and uncertainty quantification — consistent with a medium reproducibility readiness rating and a reproducibility score of 55, the mean over one reviewed paper. The bibliometric snapshot is narrow but sharply concentrated: all cataloged output falls in a single field, machine learning systems (synthetic data generation for LLMs), producing a focus score of 100, alongside 10 distinct coauthors and a collaboration score of 48. Output, impact, and activity remain unscored because the underlying data is insufficient. Name-resolved from public bibliographic data, this is an early-stage profile expected to sharpen as more work enters the catalog.

Collaboration pattern

Greco's collaboration footprint is compact: 10 distinct coauthors, with a stable core including Nabin Mulepati, Andre Manoel, Eric Tramel, Kirit Thadaka, Yev Meyer, and Alex Watson. The collaboration score of 48 reflects a concentrated, team-style working pattern typical of product-adjacent ML systems groups rather than a wide cross-institutional network. All recorded co-authorship ties to the single cataloged 2026 paper.

Research keywords

synthetic data generationLLM training datamultimodal datasetsdeclarative pipelinesdata-centric AImachine learning systemsopen-source frameworksNemotron

Highlights

First-authored open-source framework for multimodal synthetic data
As first author of NeMo Data Designer (2026, arXiv), Greco led an extensible framework that generates multimodal synthetic training data for LLMs through declarative pipelines, positioned within the Nemotron ecosystem.
Real case studies anchor the framework
Evolve's review characterizes NeMo Data Designer as a well-engineered open-source framework backed by real case studies, indicating the design is exercised on genuine data-generation workflows.
Reported gains lack ablations and uncertainty
The same review flags that reported gains lack ablation studies and uncertainty quantification, which pairs with a medium reproducibility readiness rating and drives the reproducibility dimension score of 55.

Timeline

2026First-authored NeMo Data Designer appears on arXiv, introducing an extensible framework for multimodal synthetic data generation.
2026Evolve reviews the paper as a well-engineered open-source framework backed by real case studies, while noting gains lack ablations and uncertainty.
2026Bibliometric snapshot records 10 distinct coauthors and output concentrated in a single field, yielding a focus score of 100.

Papers in the Evolve catalog

2026-09-17first author
NeMo Data Designer: An Extensible Framework for Multimodal Synthetic Data Generation

Frequent co-authors

Nabin Mulepati×1 Andre Manoel×1 Eric Tramel×1 Kirit Thadaka×1 Mike Knepper×1 Dhruv Nathawani×1 Dane Corneil×1 Yev Meyer×1 Alex Watson×1 Maarten Van Segbroeck×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.

Bibliometric scores are computed from OpenAlex-derived data only; other databases may add records.
The catalog currently contains one reviewed paper, so every scored dimension rests on a very small base.
No institution, country, or ORCID was available; identity is name-resolved from public bibliographic data, so homonyms are possible.
Output, impact, and activity axes are unscored due to insufficient data.