Generative modeling researcher: guided flow models and protein sequence-structure co-design (Université de Montréal)
2
papers here
2
reviewed by Evolve
0
first-author
0
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 input data lacks an indexed publication count, venue list, or productivity trend. This axis would measure the volume and consistency of research output over time.
Impact—
Not yet scored: no citation counts, h-index, or citation trajectory is available in the provided data. This axis would measure influence — how widely the work is cited and adopted over time.
Focus0/100
Focus scores 0/100, reflecting the most dispersed profile in this framework: the reviewed output splits exactly evenly (0.5/0.5) between machine learning for generative models and machine learning for computational biology (protein design). The two 2026 papers — one on guidance for flow-based language generation, one on protein sequence-and-structure co-design — sit in different communities with different evaluation norms, indicating deliberate breadth across generative modeling rather than concentration in a single specialty.
Collaboration50/100
Collaboration scores 50/100, the midpoint of the scale, based on 11 distinct coauthors in the bibliometric snapshot. The recurring partner list includes Jiatao Gu, Navdeep Jaitly, Joshua M. Susskind, Yizhe Zhang, Yuyang Wang, Rohit Dilip, Tianrong Chen, Jiarui Lu, and David Van Valen — a moderately sized, stable network consistent with repeated small-team work rather than broad ad-hoc collaboration.
Activity—
Not yet scored: the data provides no publication-year series beyond the two 2026 papers, so career-wide cadence cannot be measured. This axis would track publishing frequency and consistency across the full record.
Reproducibility55/100
Reproducibility scores 55/100, the mean across 2 Evolve-assessed papers, each rated medium readiness. 'How to Guide Your Language Flow' reports credible but single-seed gains with no stated code release; 'SimpleDesign' is judged honest and well-ablated but statistically thin. The pattern suggests good methodological transparency with remaining gaps in statistical rigor and artifact availability.
Research summary
Miguel Angel Bautista is profiled here from public bibliographic data and two Evolve-reviewed papers, both arXiv preprints dated 2026 on which he is the last author (non-corresponding) — a position typically associated with senior or supervisory roles. The reviewed output splits exactly evenly (0.5/0.5) between machine learning for generative models and machine learning for computational biology (protein design), which also explains his low focus score (0/100): the work spans two directions rather than one narrow lane. 'How to Guide Your Language Flow' shows MLP-probe guidance on frozen diffusion-language-model states rivaling autoguidance at far lower cost, with gains judged credible but single-seed. 'SimpleDesign' proposes a tokenizer-free, single-stage joint protein sequence-and-structure co-design model, competitive with tokenized protein language models though trailing geometric baselines; Evolve judges it honest and well-ablated but statistically thin. Reproducibility scores 55/100, the mean across 2 assessed papers, both rated medium readiness. Collaboration scores 50/100 with 11 distinct coauthors; recurring partners include Jiatao Gu, Navdeep Jaitly, Joshua M. Susskind, Yizhe Zhang, Yuyang Wang, Rohit Dilip, and David Van Valen. Output, impact, and activity axes remain unscored because the underlying bibliometric data is insufficient.
Collaboration pattern
The bibliometric snapshot lists 11 distinct coauthors, with a recurring core: Jiatao Gu, Navdeep Jaitly, Joshua M. Susskind (appearing under two name variants), Yuyang Wang (listed twice), Yizhe Zhang, Rohit Dilip, Tianrong Chen, Jiarui Lu, and David Van Valen. Both 2026 papers place Bautista in the last author position (non-corresponding), consistent with senior, supervisory involvement in small, stable teams spanning generative modeling and protein design.
Research keywords
Highlights
Cheaper guidance via probes on frozen language-model states
Tokenizer-free joint protein sequence-and-structure co-design
Timeline
- 2026Last-author arXiv preprint 'How to Guide Your Language Flow' appears, introducing MLP-probe guidance for flow-based language generation.
- 2026Last-author arXiv preprint 'SimpleDesign' appears, presenting a single-stage, tokenizer-free joint protein sequence-and-structure co-design model.
- 2026Reviewed output splits evenly (0.5/0.5) between generative models and protein design, with 11 distinct coauthors recorded in the bibliometric snapshot.
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 and radar scores are derived from Evolve's OpenAlex-indexed public bibliometric snapshot only; other indexes may differ.
- Both reviewed papers are 2026 arXiv preprints; peer-review status is not recorded in the data.
- No ORCID or country is provided; the profile is name-resolved from public bibliographic data, and 'Miguel Angel Bautista' is a common name — verify identity before reuse.
- Output, impact, and activity are unscored: no publication counts, citation metrics, or multi-year record are available in the input.