Arunava Nag

Email: arunava.nag@bsd.uchicago.edu Phone: 7754378668 Linkedin: arunavanag Github: arunavanag591

Professional Summary

Computational biologist with a strong foundation in quantitative methods and engineering, specializing in computational pathology and multimodal spatial biology — CODEX spatial proteomics, Xenium spatial transcriptomics, and H&E imaging. My work focuses on immunology and inflammatory kidney disease, applying machine learning to high-dimensional imaging data and developing reproducible Python pipelines for annotation, segmentation, statistical modeling, and simulation to reconstruct disease trajectories in pseudotime.

Work Experience

Postdoctoral Scholar, University of Chicago

May 2025 – Current

Develop scalable analysis workflows for a 68-plex CODEX lupus nephritis cohort spanning ~80 patients, 269 kidney biopsies, and more than 1 million cells, integrating Xenium spatial transcriptomics and H&E histopathology data.
Built a transformer-based cell-phenotyping model that achieved 91% accuracy against pathologist-annotated ground truth, enabling quantitative analysis of cellular composition and spatial organization in kidney tissue.
Build reproducible, Git-versioned pipelines for cell segmentation, phenotyping, spatial-neighborhood analysis, and tissue-network modeling; evaluate DINOv3, TIPSv2, and ResNet for histopathology feature extraction.
Develop H&E–CODEX image-registration methods and collaborate with nephrologists and renal pathologists to identify disease-associated cellular states, spatial biomarkers, and immune-microenvironment features linked to renal outcomes. Skills: CODEX, Spatial Omics, machine learning.

Research Assistant, University of Nevada, Reno

Jan 2020 – May 2025

Developed predictive and interpretable models for 15 million rows of noisy, multi-sensor time-series data using regression, Bayesian methods, machine learning, and Kalman-filter state-space modeling, achieving 82% prediction accuracy.
Designed open-source probabilistic simulators using autoregressive, Gaussian-process, and Bayesian-optimization methods to generate realistic time series for reproducible, large-scale experiments (open-source: COSMOS).
Integrated an odor simulator with a physics-enabled robotics simulator to build a framework for designing UAVs for outdoor odor-tracking challenges.

Senior Research Engineer, ROS-I AP, ARTC

Sep 2017 – Dec 2019

Built computer-vision / deep-learning (YOLO) and virtual-reality tools that mapped virtual to real-world space and delivered intuitive interfaces for non-technical operators, improving training efficiency 85%. Skills: Robotics, computer vision, deep learning, YOLO, virtual reality, Python, C++, C#.

Technical Skills

Machine Learning & Statistics: PyTorch, TensorFlow, CUDA, vision foundation models (DINOv3, TIPSv2), CNNs (U-Net, ResNet), Bayesian inference and optimization, Kalman filtering, time-series modeling.
Computational Pathology & Image Analysis: CODEX spatial proteomics, H&E histopathology, cross-modality image registration, affine transformations, cell segmentation and phenotyping (Cellpose), spatial-neighborhood analysis, tissue-network analysis.
Spatial & Single-Cell Biology: Xenium spatial transcriptomics, single-cell RNA-seq, Scanpy, scGPT embeddings, immune-microenvironment analysis, spatial biomarker discovery.
Programming & Scientific Computing: Python, C++, NumPy, pandas, PySpark, SciPy.
Reproducibility & Data Tools: Git.

Education

University of Nevada, Reno — PhD, Computer Science (May 2025)3.8/4.0
North Carolina State University — MS, Electrical Engineering (2016)3.2/4.0
Visvesvaraya Technological University — BE, ECE (2013)3.5/4.0

Relevant Publications

Hara S, Ai J, Cao T, Torcasso M, Walsh B, Nag A, Andrade M, Chang A, Clark M. “Spatial immune-cell profiling detects a disease-specific immune microenvironment in IgG4-related kidney disease.” Kidney International Reports (2026). [Conference abstract, World Congress of Nephrology]
Nag Arunava, and Floris Van Breugel. “COSMOS: A Data-Driven Approach to Simulating Odor Encounters for Agents Moving Through Chemical Plumes of Various Scales”, IEEE Open Access (2025)
Nag Arunava, and Floris Van Breugel. “Odour source distance is predictable from a time history of odour statistics for large scale outdoor plumes”, Journal of Royal Society Interface (2024)
Lingenfelter Bryson, Arunava Nag, and Floris van Breugel. “Insect inspired vision-based velocity estimation through spatial pooling of optic flow during linear motion.” Bioinspiration & Biomimetics (2021)

Relevant Projects

GPT based time series prediction for real world data

Replaced TIMEGAN (Time-series Generative Adversarial Networks) with GPT models, resulting in an 84% improvement in prediction accuracy for spatial forecasting of time series data from dynamic chemical sensors.

Single cell CAR T response modeling

Modeled ~18,000-gene single-cell RNA-seq profiles using Scanpy, random forests, and convolutional neural networks, and applied scGPT embeddings to predict treatment response and infer gene–gene interactions (81% accuracy).

LLM based video summarizer

Made a fast online video summarizer using the Gemini Pro LLM model and the YouTube video API.

Data Driven Plume Simulator

A data-driven fast chemical plume simulator was developed to provide a realistic experience for a simulated agent, using gaussian process and Bayesian optimization methods (PyTorch and Python).

Relevant Coursework

Statistical Analysis: Applied Regression Analysis, Advanced Probability (Bayesian Networks)
Machine Learning: Deep Learning, Reinforcement Learning, Application of Graph Theory
Data-Driven Dynamic System Modeling: Control Systems

Presentations

Invited for participating and speaking at BISCCIT, Bio-Inspired Sensing, Computing, and Control with International Teams, 2023, London.
Presented “Outdoor odor localization” and “Data driven plume simulator” at the AI Dynamics Institute workshop, University of Washington, 2023.
Presented “Mapping outdoor odor plumes using a mobile chemical sensor” at APS 2022, Chicago.