Doctoral Researcher · AI & Machine Learning · Healthcare

Bridging math,
machines & medicine.

Developing interpretable AI-based healthcare solutions particularly for resource-constrained settings of developing nations in collaboration with AIIMS-Delhi.

IIT Delhi  ·  IBM Research  ·  NIT Hamirpur

Portrait of Aman Bhardwaj

About

Who I am

Researcher working at the intersection of mathematics, machine learning, and medicine to develop interpretable, physiology-informed AI for healthcare in resource-constrained settings.

I develop constrained optimization methods that build domain knowledge directly into computational models, rather than treating them as black boxes. Together with clinicians at AIIMS-Delhi and AFMC (Armed Forces Medical College), Pune, I work on diagnostic tools designed to be transparent and usable where specialist care is scarce.

At IBM India Research Lab, I apply the same emphasis on reliability to enterprise AI, building LLM-based systems for modernizing legacy code in financial and healthcare domains, including work on retrieval-augmented generation (RAG) and AI agents.

I'm always happy to talk with researchers and clinicians working on similar problems.

Research Focus

What I work on

Three interlocking threads — physiology, clinical AI, and enterprise systems — that together define my research identity.

Physiology-Informed AI

Developing physiology-informed models for interpretable ECG analysis and holistic cardiac profiling from long-term recordings.

Enterprise AI Systems

Architecting spec-driven AI-workflows for legacy enterprise-scale system modernization, garnering skills in LLMs, RAG, and AI Agents.

Brain Stroke Diagnosis

Robust ML frameworks for early stroke diagnosis in resource-limited environments, addressing the 3rd leading cause of disability globally.

Education

Academic path

PhD in Computer Science

Indian Institute of Technology Delhi (IIT-Delhi)

Oct 2022 – Present

  • CGPA: 9.43/10
  • Research Focus: Physiology-informed Representation Learning from Electrocardiograms
  • Advisors: Prof. Rahul Garg & Prof. Srikanta Bedathur (CSE, IIT-Delhi), Dr. Vishnu VY (AIIMS-Delhi)
  • Clinical Collaborators: AIIMS-Delhi & INHS Aswini-Mumbai

MS (Research) in Computer Science

Indian Institute of Technology Delhi (IIT-Delhi)

Jan 2020 – Jun 2022

  • CGPA: 9.68/10
  • Thesis: Machine Learning Based Stroke Detection and Stroke Type Identification in Resource Limitations
  • Advisor: Prof. Rahul Garg | Clinical Co-investigator: Dr. VY Vishnu (AIIMS-Delhi) and Dr. Vinny Wilson (AFMC-Pune)

B.Tech. in Electronics and Communication Engineering

National Institute of Technology Hamirpur (NIT-Hamirpur)

2012 – 2016

  • CGPA: 8.73/10
  • President, Society for Promotion of Electronics Culture (SPEC)
  • Placement Coordinator

Selected Publications

Research output

Refining LLM-Based COBOL-to-Java Translation via Natural Language Summary Augmentation

33rd IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER) 2026

A. Bhardwaj, V. Arya, Y. Sabharwal.

Oral Presentation · Limassol, Cyprus

A novel machine learning framework for stroke type identification in resource constrained settings with robustness to missing data

Nature Scientific Reports, 2025

A. Bhardwaj, Y Antil, MVP Srivastava, PW Vinny, VY Vishnu, R Garg

Impact Factor: 4.3

Calibrating machine learning models for accurate stroke type prediction in low resource settings

International Journal of Stroke, 2024

A Mehta, A Bhardwaj, Y Antil, MVP Srivastava, VP Wilson, V Vishnu, R Garg

Impact Factor: 8.7

Lost In The Past: How domain shift in data impacts the performance of ML-based models for Stroke type identification

Journal of the Neurological Sciences, World Congress of Neurology (WCN) 2023

A. Bhardwaj, Y Antil, MVP Srivastava, PW Vinny, VY Vishnu, R Garg

Poster Presentation · Montreal, Canada · Impact Factor: 3.2

A View Independent Classification Framework for Yoga Postures

Springer Nature Computer Science, 2022

M Chasmai, N Das, A. Bhardwaj, R Garg

Machine Learning Based Reanalysis of Clinical Scores for Distinguishing Between Ischemic and Hemorrhagic Stroke in Low Resource Setting

Journal of Stroke and Cerebrovascular Diseases, 2022

A. Bhardwaj, MVP Srivastava, PV Wilson, A Mehndiratta, VY Vishnu, R Garg

Poster Presentation · Montreal, Canada · Impact Factor: 2.67

Protocol for visual data collection to enable Computer Vision and Deep Learning for the identification of Stroke and Stroke-type in resource limitations

Journal of the Neurological Sciences, World Congress of Neurology (WCN) 2023

A. Bhardwaj, Y Antil, MVP Srivastava, PW Vinny, VY Vishnu, R Garg

Oral Presentation · Montreal, Canada

Clinical Identification Of Stroke And Stroke Type Via Clinical Attributes And Gait Analysis In Resource Limited Environment

International Journal of Stroke, World Stroke Congress 2022

A. Bhardwaj, Y Antil, MVP Srivastava, PW Vinny, VY Vishnu, R Garg

Oral Communication · Impact Factor: 6.948

Research Projects

Building blocks

Physiology-informed Models

PhD Research · Ongoing

Constrained optimization algorithm integrating finite cardiac conduction velocities for interpretable ECG alignment. Matches deep learning performance while maintaining complete clinical interpretability.

Optimization Healthcare AI Interpretability

Wearable Biomarker Monitoring

Ongoing Research

Non-invasive continuous monitoring of blood glucose and electrolytes via wearable devices for cardio-cerebrovascular risk assessment.

Wearables Signal Processing Healthcare

LLM-based Legacy System Modernization

IBM Research

Transformer-based models for neural machine translation and automated code generation across programming languages with RAG and agentic AI workflows.

LLMs RAG AI Agents WCA4Z

Early Stroke Diagnosis System

Master's Research

Robust ML framework for stroke type identification handling missing data and class imbalance. Enables preliminary diagnosis without neuroimaging in resource-constrained environments.

Machine Learning Clinical AI AIIMS Collaboration

Oblique Sparse Factor Models

Research Tool

Generalized Oblique Sparse Factor Model — doubly-penalized matrix decomposition for exploratory analysis with superior sparsity and interpretability on small datasets.

Matrix Decomposition Sparse Models Interpretability

Computer Vision for Stroke Screening

Clinical Collaboration

10-test audio-visual protocol combining NIHSS-8 and Yoga-inspired postures for deep learning gait analysis. Piloted at AIIMS-Delhi.

Computer Vision Deep Learning Clinical Validation

Updates & Recognition

News, awards & service

A running log of recent milestones — from paper acceptances and conference presentations to grants and community service.

Mar 2026
News

Oral Presentation at SANER 2026, Limassol, Cyprus

Presented "Refining LLM-based COBOL-to-Java Translation via Natural Language Summary Augmentation".

Dec 2025
News

Paper Accepted at SANER 2026

First industrial paper at IBM Research in collaboration with Vijay Arya and Yogish Sabharwal.

Oct 2025
News

Paper Accepted at Nature Scientific Reports

A Novel Machine Learning Framework for Stroke Type Identification in Resource Constrained Settings with Robustness to Missing Data.

Sep 2025
Service

Program Committee Member — CODS-COMAD 2025

Selected as PC Member (Applied Data Science Track) for ACM 13th International Conference on Data Science.

Jul 2024
News

Joined IBM Research

Started as Research Engineer at IBM India Research Lab, AI4Code Team, working on LLMs and enterprise AI systems.

Mar 2024
Award

SERB Core Research Grant

Awarded Core Research Grant by Science and Engineering Research Board (SERB), DST, Government of India for prospective diagnostic study on wearable devices in acute stroke patients.

Dec 2023
Service

Google Research Week 2023

Selected for Google Research Week 2023 in Bengaluru, featuring workshops and keynote sessions on machine learning foundations.

Jun 2023
News

WCN 2023 — Two Abstracts Accepted

One poster and one oral presentation accepted at World Congress of Neurology (WCN) 2023, Montreal, Canada.

Jun 2023
News

PhD Coursework Completed

Completed PhD coursework with CGPA 9.43/10 at IIT Delhi.

2022
Award

PhD Fellowship — IIT Delhi

Institute PhD Fellowship from Indian Institute of Technology Delhi.

2020
Award

National-Level Winner — ICVGIP 2020

Visual Data Challenge, Indian Conference on Computer Vision, Graphics and Image Processing (ICVGIP) 2020.

2020
Award

MS Research Fellowship — IIT Delhi

Institute MS Research Fellowship from Indian Institute of Technology Delhi.

2009
Award

All India Certificate of Merit, Science — AISSE CBSE

Ranked in the top percentile nationally in the All India Secondary School Examination.

Get In Touch

Let's connect

Always interested in discussing research collaborations, healthcare AI projects, or opportunities in interpretable machine learning. Feel free to reach out.

Location

Room 409, SIT-Building, IIT Delhi, New Delhi, India