CV

Education

Ph.D., MIT School of Engineering, 2025
- Advisor: Prof. Josh McDermott, Laboratory for Computational Audition
- Thesis: Synthesis and perception of sounds from physical interactions reveals auditory intuitive physics
B.Tech. and M.Tech., Mechanical Engineering, IIT Bombay, 2019
- Focus: Acoustics and automation
- Masters thesis: Inverse acoustic characterization of periodic reticulated foam (with Prof. Sripriya Ramamoorthy)

Work Experience

Principal AI Engineer, K-Dense Inc., 2024 - Present
- Lead AI research on transcriptomic aging, foundation models for biology, and autonomous scientific analysis
- Built ensemble transcriptomic aging clock: R2 = 0.854, MAE = 4.26 years, trained on 57,584 samples across 28 tissues
- Co-developed GeneJepa (predictive world model of the transcriptome) and Revive-Flow (foundation model for blood DNA methylation aging)
- Built K-Dense Analyst: autonomous AI system for end-to-end scientific analysis
- K-Dense is backed by Google AI Futures Fund and Accel Atoms program
Research Intern, Meta Reality Labs Research, Summer 2023
- Augmented spatial audio rendering on smart glasses; developed models for online sound synthesis
Graduate Research Assistant, MIT Laboratory for Computational Audition, 2019 - 2025
- Studied auditory intuitive physics: rigid-body impact, scraping, and rolling sounds
- Built differentiable physical synthesis models; designed and ran psychophysical experiments
- Advisors: Prof. Josh McDermott and Prof. Josh Traer
Research Intern, CNRS Laboratoire de Mecanique et d’Acoustique, Marseille, France
- SoniMove project: perceptual studies and sound synthesis
Research Intern, Simon Fraser University
- Automatic song transcription (with Prof. Arne Eigenfeldt)

Skills

Machine learning: PyTorch, JAX, transformers, diffusion models, probabilistic modeling
Computational biology: transcriptomics, epigenomics, aging clocks, RNA-seq pipelines
Scientific computing: Python, NumPy, SciPy, Pandas, scikit-learn
Audio and acoustics: physical synthesis models, psychoacoustics, signal processing
Autonomous AI systems: multi-agent pipelines, AI-driven research workflows

Publications

MetaMuse: A Multi-Agent AI System for Biomedical Metadata Curation and Harmonization

Mittal, E., Litman, E., Myers, T., Agarwal, V., Gopinath, A., & Kassis, T. (2026). MetaMuse: A Multi-Agent AI System for Biomedical Metadata Curation and Harmonization. bioRxiv. https://doi.org/10.64898/2026.04.12.718044

Limitations of TabPFN for High-Dimensional RNA-seq Analysis

Agarwal, V., Li, O., Kassis, T. et al. (2025). Limitations of TabPFN for High-Dimensional RNA-seq Analysis. bioRxiv.

Transcriptomic Age Prediction Using Mixture-of-Experts Models Reveals Tissue-Specific Aging Signatures

Agarwal, V., Li, O., Kassis, T., Gopinath, A. et al. (2025). Transcriptomic Age Prediction Using Mixture-of-Experts Models Reveals Tissue-Specific Aging Signatures. medRxiv.

K-Dense Analyst: Towards Fully Automated Scientific Analysis

Agarwal, V., Kassis, T., Li, O., Gopinath, A. et al. (2025). K-Dense Analyst: Towards Fully Automated Scientific Analysis. arXiv.

Revive-Flow: A Foundation Model for Blood DNA Methylation Aging

Agarwal, V., Li, O., Kassis, T., Gopinath, A. et al. (2025). Revive-Flow: A Foundation Model for Blood DNA Methylation Aging. bioRxiv.

GeneJepa: A Predictive World Model of the Transcriptome

Agarwal, V., Li, O., Kassis, T., Gopinath, A. et al. (2025). GeneJepa: A Predictive World Model of the Transcriptome. bioRxiv.

Object-based synthesis of scraping and rolling sounds based on non-linear physical constraints

Agarwal, V., Cusimano, M., Traer, J., & McDermott, J. H. (2021). Object-based synthesis of scraping and rolling sounds based on non-linear physical constraints. Proceedings of the 24th International Conference on Digital Audio Effects (DAFx20in21), 136–143.

Selected Awards and Affiliations

Google AI Futures Fund (K-Dense), 2024-2025
Accel Atoms Program (K-Dense), 2025

Vinayak Agarwal

CV