cv

General Information

Full Name Gabriel Miranda de Araujo
Email gabrielmidar@gmail.com
Languages Portuguese (Native), English (Fluent), Spanish (Fluent)
Programming Languages Python, Rust, Bash, C, C++, Java, R
Technical Skills Machine Learning, Computational Biology, Computer Vision, PyTorch, JAX, Linux, CLI, SLURM, TensorFlow, Keras

Education

  • 2022 - 2024
    MSc in Computer Science
    University of California, Irvine
    • GPA: 3.96/4.0
  • 2017 - 2021
    BSc in Computer Science
    University of São Paulo
    • GPA: 8.0/10.0
    • Awarded Undergraduate Research Fellowship from Brazilian National Council for Scientific and Technological Development (CNPq)

Experience

  • 2024 - Pres.
    Computational Postgraduate Associate
    Yale School of Medicine
    • Supervised by PI: Hattie Chung, PhD
    • Developed a novel algorithm, MIND, a geometric deep learning framework that infers single-cell metabolic rates and spatial metabolic "divisions of labor" using transcriptomics and spatial coordinates; incorporates Kirchhoff's law to enforce energy-balanced reaction constraints. First author paper in preparation.
    • Trained MIND on 1 million spatial spots across human and mouse tissues (brain, liver, heart, ovary), achieving >1000x speedup over existing methods (Compass).
    • Developed a novel graph-based metric for quantifying cell-cell metabolic cooperation, identifying spatially organized tasks shared across cells.
    • Co-developing a multi-instance learning (MIL) framework to integrate MOVAT, Sirius Red, and H&E whole-slide images (150 patients). Aligning >5M patches across stains using graph optimal transport and pathology foundation models (CONCH, UNI).
    • Contributed to lab projects through analysis of bulk, single-cell RNA-seq and spatial transcriptomics (10X Visium, Slide-seqV2, Stereo-seq) datasets.
    • Led bi-weekly single-cell machine learning methods subgroup. Discussions involved methods for trajectory inference, tissue motif detection and multi-omics integration architectures.
    • Mentored 5 undergraduate lab members on single-cell analysis techniques, Pytorch-based neural networks, Bash, HPC and SLURM basic usage.
  • 2023 - 2024
    Master's Graduate Student Researcher
    Donald Bren School of Information & Computer Science
    • Supervised by PI: Xiaohui Xie, PhD
    • Co-first author of CoMA, a multi-modal generative model integrating masked transformers, VQ-VAEs, and language agents for controllable 3D human motion generation. Accepted at AAAI 2026.
    • Developed a deep learning method for registering 576 T1-weighted 3D volumetric brain MRI scans using vision transformers and gradient-based optimization.
    • Mentored 4 undergraduate exchange students in multi-modal agent integration, VQ-VAEs, bash usage, and masked-transformer models.
  • 2020 - 2021
    Undergraduate Researcher Senior thesis
    Institute of Mathematics and Statistics (IME) - USP
    • Supervised by PI: Paulo Miranda, PhD
    • Developed a graph-based computer vision algorithm for medical image segmentation from MRIs, e.g. anatomical features in cross sections of the wrist, implemented in C/C++ with OpenMP parallelization.
    • Applied graph-based Unsupervised Oriented Image Foresting Transform and Min-tree data structure to create and modify super-pixel hierarchy graphs in images.
  • 2019 - 2020
    Undergraduate Researcher
    Signal Processing Laboratory (LPS) - USP
    • Supervised by PI: Hae Yong Kim, Ph.D
    • Evaluated the transfer learning performance of breast cancer detection CNN model from mammogram images trained older patients to 135 orthogonal younger patients (<40y.o) from our partner hospital (ICESP).
    • Led weekly discussion with oncologists and radiologists for qualitative model performance evaluation.
  • 2018 - 2019
    Undergraduate Researcher
    Institute of Mathematics and Statistics (IME) - USP
    • Supervised by PI: Roberto Hirata, PhD
    • Developed CNN and custom dataset (1.6K images) for urban electrical wiring detection using GeoPandas, Django, Keras and TensorFlow.

Publications and Conference Proceedings

  • Gabriel De Araujo, David van Dijk, Purushottam Dixit, Hattie Chung. MIND: Metabolic Inference of Niches through graph Diffusion. In preparation.
  • Shanlin Sun*, Jiaqi Xu*, Gabriel De Araujo*, Shenghan Zhou*, Hanwen Zhang, Ziheng Huang, Chenyu You, Xiaohui Xie. CoMA: Compositional Human Motion Generation with Multi-modal Agents. AAAI 2026.
  • Gabriel De Araujo, Shanlin Sun, Xiaohui Xie. Adaptive Image Registration: A Hybrid Approach Integrating Deep Learning and Optimization Functions for Enhanced Precision. arXiv:2311.15497 (2023).
  • Daniel Gustavo Pellacani Petrini, Gabriel Vansuita Valente, Carlos Shimizu, Rosimeire Aparecida Roela, Gabriel Miranda de Araújo, Tatiana Cardoso de Mello Tucunduva, Maria A. A. Koike Folgueira, and Hae Yong Kim. Evaluation of an AI system for breast cancer screening in mammograms of young women Journal of Clinical Oncology (2020).

Selected Talks and Posters

  • Quantiative Biology/Physics, Engineering and Biology Symposium
    Poster presentation on MIND. Hosted by the Program in Physics, Engineering and Biology (PEB). (April 2026)
  • Yale Cardiovascular Research Center - Research In Progress Talks Series
    Oral presentation on MIND to the Internal Medicine Department. Hosted by the YCVRC. (April 2026)
  • Yale Research Synergy Forum on Cardiovascular Disease and Inflammation
    Oral presentation on MIND to 50 PIs. Hosted by the Yale School of Medicine Dean's Office. (August 2025)
  • Program in Physics, Engineering and Biology (PEB) Seminar
    Oral presentation on MIND to graduate students, postdocs, and faculty. (February 2025)
  • InterSCity Workshop, USP
    Oral presentation on wire detection CNN model to graduate students, postdocs, and faculty. (February 2019)