Computational Biology · Machine Learning
Arrio Gonsalves

Using machine learning and genomic data to understand how diseases arise.

M.S. student in Computational Biology at Carnegie Mellon University, building multi-agent LLM pipelines to uncover mechanistic links between comorbid diseases. Previously applied computer vision to medical imaging at UT Southwestern and interned on AI automation for patient support at Novartis.

About

I'm a graduate researcher at Carnegie Mellon University, where I'm designing a multi-agent LLM pipeline that combines gene-disease association data and protein-protein interaction networks to construct and explain mechanistic pathways linking comorbid diseases, such as Alzheimer's–hypercholesterolemia and type 2 diabetes–coronary artery disease.

Before CMU, I spent three years at UT Southwestern Medical Center training and benchmarking U-Net and Vision Transformer architectures to detect Alzheimer's and related neurodegenerative pathology in whole-slide brain tissue images, work that earned Best Undergraduate Student Poster at the 3rd Annual Workshop on Imaging and Data Science. I hold a B.S. in Computer Science from UT Dallas with a certificate in Data Science.

Over the summer of 2026, I interned as an AI Automation Intern at Novartis Pharmaceuticals, building automated workflows to help patients receive and adhere to prescribed medications.

Education

Carnegie Mellon University
M.S. in Computational Biology, Pittsburgh, PA

Relevant coursework: Machine Learning, Quantitative Genetics, Applied Cellular and Molecular Biology, Essential Mathematics and Statistics.

The University of Texas at Dallas
B.S. in Computer Science, Richardson, TX

Undergraduate Certificate in Data Science.

Research Experience

Graduate Student Researcher — Carnegie Mellon University
Advised by Dr. Min Xu · Pittsburgh, PA

Thesis Committee: Dr. Martin Jinye Zhang, Dr. Zheng Kuang

  • Designing a multi-agent LLM pipeline that integrates public gene-disease association data (Open Targets, GWAS Catalog) with protein-protein interaction networks (OmniPath) to construct and explain mechanistic pathways linking pairs of comorbid diseases, such as Alzheimer's disease–hypercholesterolemia and type 2 diabetes–coronary artery disease.
  • Building an automated benchmark that scores candidate disease mechanisms on protein recall, hallucination rate, and interaction-directionality accuracy against literature-derived gold standards, comparing the grounded multi-agent pipeline against an ungrounded LLM baseline.
  • Validating candidate protein-protein mechanisms against independent gene-disease association databases and real-world patient diagnosis co-occurrence data (MIMIC-IV) to assess biological and clinical plausibility.
Staff Associate / Undergraduate Research Fellow (Green Fellows Program) — UT Southwestern Medical Center
Advised by Dr. Satwik Rajaram · Dallas, TX
  • Trained and benchmarked U-Net (ResNet backbone) and Vision Transformer architectures for segmentation and classification of whole-slide brain tissue images, with U-Net achieving superior accuracy identifying Alzheimer's and related neurodegenerative pathology.
  • Continued as Staff Associate following the fellowship, extending the pipeline to additional disease cohorts and refining model performance for production use.
  • Awarded Best Undergraduate Student Poster at the 3rd Annual Workshop on Imaging and Data Science (UT Dallas + UTSW).
Undergraduate Student Researcher — UT Dallas
Advised by Dr. Weili Wu · Richardson, TX
  • Investigated the feasibility of HIPAA-compliant decentralized storage and sharing of electronic health records using distributed-ledger technology, identifying scalability risks and compliance gaps.
  • Authored a literature review evaluating existing decentralized EHR frameworks and proposing directions for compliant, scalable architectures.

Industry Experience

AI Automation Intern — Novartis Pharmaceuticals Corporation
via Magnit, LLC · East Hanover, NJ
  • Designed automated, AI-driven workflows to support patients in receiving and adhering to prescribed medications, streamlining patient support operations.
  • Developed a proof-of-concept LLM-based system to automatically profile cancer patients based on their familiarity with and confidence in their prescribed therapy.
  • Transformed a theoretical concept into a fully functional proof-of-concept workflow and developed a roadmap for its transition to a production-ready system.

Projects

Chest X-Ray Diagnosis Network

Artificial Intelligence Society, UT Dallas · Feb–Apr 2023

Built and trained a neural network with Python, PyTorch, and FastAI to detect pathological conditions and infer demographic attributes from chest X-ray images.

reMember

ACM Projects Team · Aug–Nov 2021

Selected from 500+ applicants to build an Android app helping Alzheimer's patients recognize family and friends via facial recognition, using TensorFlow and Firebase for real-time inference.

Autonomous Solar Drone Flight Software

RAD Team, AIAA UT Dallas · Sep–Nov 2022

Selected from 70+ applicants to develop flight control software for an autonomous solar-powered drone, programming takeoff, flight, and landing sequences.

InterReview — 1st Place, HackUTD

RingCentral Sponsor Challenge · Nov 2022

Co-developed a web app to help users practice public speaking and interview skills; built core backend infrastructure with Firebase.

Skills & Certifications

Programming

Python Java C++ SQL

ML & Computational

PyTorch TensorFlow U-Net & Vision Transformers LLM-based Pipelines FastAI Scikit-learn Pandas NumPy

Tools & Methods

Git Jupyter Whole-Slide Image Analysis Literature Review Research Writing Agile Development

Certifications

Supervised ML: Regression & Classification (DeepLearning.AI) AI Foundations Specialist (IBM) Introduction to AI (IBM) Getting Started with AI using IBM Watson MTA: Intro to Programming Using Java

Honors & Awards

  • Best Undergraduate Student Poster, 3rd Annual Workshop on Imaging and Data Science (UT Dallas + UTSW)Oct 2024
  • President's Office Award, UT Dallas Summer Platform of Undergraduate Research (SPUR)Aug 2024
  • UT Dallas – UTSW Green FellowshipSpring 2024
  • UT Dallas Dean's ListFall 2023 & Spring 2024
  • HackUTD RingCentral Sponsor Challenge — 1st Place WinnerNov 2022
  • HackAI Brainiac Challenge — 1st Runner-Up
  • UT Dallas Academic Excellence Scholarship (all semesters)Fall 2021 – Spring 2025