Evgeny Burnaev

—

Unclaimed · AI-structured profileOpenAlex

Machine learning researcher bridging generative diffusion imputation and topological analysis of code LLMs

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—

Output is not scored: this snapshot provides no publication counts, venue list, or full paper record, so productivity cannot be measured. This axis would track the volume and cadence of peer-reviewed and preprint outputs over time.

Impact—

Impact is not scored: no citation counts, h-index, or venue prestige data were provided in this snapshot. This axis would measure how widely the work is cited and taken up by other researchers across the community.

Focus0/100

The focus score is 0, reflecting an even 50/50 split of the cataloged record between machine learning for software engineering (code LLMs) and machine learning for generative models and data imputation. The two 2026 arXiv papers, Impute-EM and CodeTD, sit one in each field, so the visible record shows breadth across two threads rather than concentration in a single niche; a larger catalog would be needed to tell whether one thread dominates.

Collaboration46/100

Collaboration scores 46, built on 9 distinct coauthors in the cataloged record. Recurring partners include Daria Voronkova, Ilya Trofimov, Anton Dmitriev, Eduard Tulchinskii, Serguei Barannikov, Sergei Kholkin, Kirill Sokolov, Dmitry Baranchuk, and Alexander Korotin. On both reviewed 2026 papers the researcher sits in a middle author position and is not the corresponding author, indicating contribution within larger team-written arXiv papers rather than solo or lead-authored work.

Activity—

Activity is not scored: the snapshot lacks a full publication timeline, so recent publishing cadence cannot be assessed. This axis would measure output recency and sustained research tempo across recent years.

Reproducibility55/100

Reproducibility scores 55, the mean over the 2 reviewed papers, both rated medium reproduction readiness. Evolve's verdicts flag concrete gaps: Impute-EM reports strong tabular fidelity gains but no error bars, and CodeTD shows broad solid experiments yet modest, mixed gains over spectral baselines. The mid-range score reflects solid experimental coverage undermined by missing uncertainty reporting and limited headroom over simpler baselines; it is coarse because only two papers are scored.

Research summary

Evgeny Burnaev is a machine learning researcher whose visible record in the Evolve catalog spans two complementary threads: generative modeling for missing-data problems and machine learning applied to software engineering, with the cataloged field distribution split evenly at 50/50. On both 2026 arXiv papers reviewed, Burnaev appears as a middle author and not the corresponding author. Impute-EM proposes EM-style imputation built on native mixed-state diffusion backbones, combining masked discrete diffusion and expectation-maximization for mixed-type tabular data; Evolve notes strong tabular fidelity gains while judging the contribution incremental over DiffEM and DiffPuter. CodeTD applies topological data analysis — computing Manifold Topology Divergence on attention maps with an XGBoost classifier — to detect incorrect code before execution; Evolve finds the experiments broad and solid but the gains over spectral baselines modest and mixed. The reproducibility dimension scores 55, the mean across the 2 reviewed papers, both rated medium reproduction readiness, with missing error bars flagged for Impute-EM. Collaboration scores 46 across 9 distinct coauthors, including Daria Voronkova, Eduard Tulchinskii, Serguei Barannikov, Dmitry Baranchuk, and Alexander Korotin, indicating steady participation in team-authored work. Output, impact, and activity are not scored because the snapshot lacks citation, venue, and full-timeline data; institution, country, and ORCID were also absent, so this profile is name-resolved from public bibliographic data only.

Collaboration pattern

Collaboration scores 46 from 9 distinct coauthors. Recurring partners named in the record include Daria Voronkova, Ilya Trofimov, Anton Dmitriev, Eduard Tulchinskii, Serguei Barannikov, Sergei Kholkin, Kirill Sokolov, Dmitry Baranchuk, and Alexander Korotin. On both 2026 papers Burnaev sits in a middle author position and is not the corresponding author, so the visible pattern is contributing within team-authored arXiv work rather than leading papers as first or corresponding author.

Research keywords

missing data imputationdiffusion modelsexpectation-maximizationmixed-type tabular datamasked discrete diffusiongenerative modelscode LLM hallucination detectiontopological data analysisattention mapsManifold Topology Divergencepre-execution code verification

Highlights

Impute-EM: EM-style diffusion imputation for mixed-type tabular data
The 2026 arXiv paper pairs expectation-maximization with native mixed-state diffusion backbones and masked discrete diffusion for heterogeneous data imputation; Evolve notes strong tabular fidelity gains, while judging the advance incremental over DiffEM and DiffPuter and flagging the absence of error bars.
CodeTD: topology of attention flags code LLM hallucinations before execution
CodeTD applies topological data analysis to attention maps — computing Manifold Topology Divergence with an XGBoost classifier — to detect incorrect code pre-execution; Evolve rates the experimental coverage broad and solid, but finds the gains over spectral baselines modest and mixed.

Timeline

2026Impute-EM appears on arXiv, proposing EM-style imputation built on native mixed-state diffusion backbones for heterogeneous data.
2026CodeTD appears on arXiv, using topological analysis of attention maps to detect hallucinations in code LLMs before execution.
2026Both cataloged papers fall in 2026, so the visible record is concentrated in a single year; earlier output is not captured in this snapshot.

Papers in the Evolve catalog

Frequent co-authors

Daria Voronkova×1 Ilya Trofimov×1 Anton Dmitriev×1 Eduard Tulchinskii×1 Serguei Barannikov×1 Sergei Kholkin×1 Kirill Sokolov×1 Dmitry Baranchuk×1 Alexander Korotin×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 dimensions are computed from OpenAlex-style bibliometric data only; no citation counts, h-index, or venue list were available in this snapshot.
Only the 2 Evolve-reviewed papers are covered here; the researcher's broader publication record is not represented.
Institution, country, and ORCID were absent from the input data; the profile is name-resolved from public bibliographic data and identity is not independently confirmed.
The reproducibility score (55) is a mean over only 2 papers, and the output, impact, and activity axes lack underlying data entirely.