Mehdi Yazdani-Jahromi

AI Research Scientist · Computational Biology

  • DeepBio Scientific
  • Microsoft
  • Johnson & Johnson
  • University of Central Florida

Representation learning for biological systems, from mRNA language models to drug–target interaction.

I’m a Senior Research Scientist at DeepBio Scientific, where I lead research on machine learning for biomedical data and agentic systems for retrieval and reasoning. Previously I was at Microsoft and Johnson & Johnson, and earned my Ph.D. in Computer Science at UCF.

View publicationsCurriculum vitae →
Mehdi Yazdani Jahromi

Selected papers

All 15 publications →
  1. HELM: Hierarchical Encoding for mRNA Language Modeling (external link)Mehdi Yazdani-Jahromi, Mangal Prakash, Tommaso Mansi, Artem Moskalev, Rui Liao
    International Conference on Learning Representations (ICLR 2025)
    Cite
    @article{yazdani2024helm,
  title={HELM: Hierarchical Encoding for mRNA Language Modeling},
  author={Yazdani-Jahromi, Mehdi and Prakash, Mangal and Mansi, Tommaso and Moskalev, Artem and Liao, Rui},
  journal={ICLR 2025},
  url={https://arxiv.org/abs/2410.12459},
  year={2024}
}
  2. Equi-mRNA: Protein Translation Equivariant Encoding for mRNA Language Models (external link)Mehdi Yazdani-Jahromi, Ali Khodabandeh Yalabadi, Ozlem Ozmen Garibay
    Advances in Neural Information Processing Systems 38, 2025
    Cite
    @inproceedings{Yazdani_Jahromi_2025, series={NeurIPS 2025}, title={Equi-mRNA: Protein Translation Equivariant Encoding for mRNA Language Models}, url={http://dx.doi.org/10.52202/085713-0904}, DOI={10.52202/085713-0904}, booktitle={Advances in Neural Information Processing Systems 38}, publisher={Neural Information Processing Systems Foundation, Inc. (NeurIPS)}, author={Yazdani-Jahromi, Mehdi and Khodabandeh Yalabadi, Ali and Garibay, Ozlem}, year={2025}, pages={30416–30456}, collection={NeurIPS 2025} }
  3. Fair Bilevel Neural Network (FairBiNN): On Balancing fairness and accuracy via Stackelberg Equilibrium (external link)Mehdi Yazdani-Jahromi, Ali Khodabandeh Yalabadi, AmirArsalan Rajabi, Aida Tayebi, Ivan Garibay, Ozlem Garibay
    Advances in Neural Information Processing Systems 37, 2024Code
    Cite
    @inproceedings{Yazdani_Jahromi_2024, series={NeurIPS 2024}, title={Fair Bilevel Neural Network (FairBiNN): On Balancing fairness and accuracy via Stackelberg Equilibrium}, url={http://dx.doi.org/10.52202/079017-3355}, DOI={10.52202/079017-3355}, booktitle={Advances in Neural Information Processing Systems 37}, publisher={Neural Information Processing Systems Foundation, Inc. (NeurIPS)}, author={Yazdani-Jahromi, Mehdi and Yalabadi, Ali and Rajabi, Amirarsalan and Tayebi, Aida and Garibay, Ivan and Garibay, Ozlem}, year={2024}, pages={105780–105818}, collection={NeurIPS 2024} }
  4. AttentionSiteDTI: an interpretable graph-based model for drug-target interaction prediction using NLP sentence-level relation classification (external link)Mehdi Yazdani-Jahromi, Niloofar Yousefi, Aida Tayebi, Elayaraja Kolanthai, Craig J Neal, Sudipta Seal, Ozlem Ozmen Garibay
    Briefings in Bioinformatics, 2022Code
    Cite
    @article{Yazdani_Jahromi_2022, title={AttentionSiteDTI: an interpretable graph-based model for drug-target interaction prediction using NLP sentence-level relation classification}, volume={23}, ISSN={1477-4054}, url={http://dx.doi.org/10.1093/bib/bbac272}, DOI={10.1093/bib/bbac272}, number={4}, journal={Briefings in Bioinformatics}, publisher={Oxford University Press (OUP)}, author={Yazdani-Jahromi, Mehdi and Yousefi, Niloofar and Tayebi, Aida and Kolanthai, Elayaraja and Neal, Craig J and Seal, Sudipta and Garibay, Ozlem Ozmen}, year={2022}, month=July }

Research

Research overview →

  1. Language models for mRNA

    Encodings that reflect codon hierarchy and protein translation, improving property prediction and sequence design.

  2. Drug–target interaction & molecule design

    Interpretable models over 3D binding sites and fragments, and diffusion-based generation of target-specific molecules.

  3. Trustworthy & agentic ML

    Fair representation learning with bilevel formulations, and agents that retrieve, reason and cite over biomedical data.

News

All news →
Oct 2026MIRAGE Selected as a Top 4 Paper at MoML 2026Our paper "MIRAGE: Measuring Interpolation and Redundancy in Affinity GEneralization" has been selected as one of the Top 4 submissions at MoML 2026 and will be presented as a flash talk at MIT.

Blog

All posts →