Academic Work

Research & Publications

Peer-reviewed papers, technical reports, and ongoing research in AI safety, healthcare, and biosecurity.

Master's Thesis

Enhancing Healthcare Insurance Fraud Detection and Prevention with a Machine Learning and Blockchain-Based Approach

Supervisor: Dr. Rajesh Kumar Sinha, Assistant Professor
Institution: National Institute of Technology Patna
Department: Department of Mathematics
Period: January - June, 2024

Abstract

This research proposes a healthcare insurance fraud detection system combining an Optuna-tuned weighted ensemble of five gradient-boosted classifiers (F1 = 0.7345, leakage-free 10-fold CV) with a SHA-256 blockchain audit trail and ECIES encryption. The system aggregates 558,211 Medicare claims into 5,410 provider-level records with 190 engineered features. SHAP and LIME provide dual explainability for fraud risk assessments.

Healthcare Fraud
Machine Learning
Blockchain
ECIES Encryption
XGBoost
SHAP
LIME
Optuna
Ensemble Models
Merkle Tree

Publications

7

Artistic Style Transfer using Generative Adversarial Networks: A Pix2Pix Implementation

Pavan Kumar Dubasi, Bhadula, S., et al.

2024 International Conference on Communication, Computer Sciences and Engineering (IC3SE)

Date of Conference 09-11 May 2024, Date Added to IEEE Xplore 23 July 2024

We propose an enhanced Pix2Pix GAN architecture for artistic style transfer that achieves superior results in preserving structural integrity while transferring complex artistic styles.

GANs
Style Transfer
Deep Learning
Computer Vision

Verifiable Identity and Audit Infrastructure for DNA Synthesis Compliance

Pavan Kumar Dubasi

BlueDot Impact Biosecurity Programme 2026

IEEE conference paper format, Feb 2026

Proposes a verifiable identity and audit infrastructure for DNA synthesis compliance using W3C decentralized identifiers and verifiable credentials. Includes independent deployment and testing of the NAO metagenomic surveillance pipeline, with discovery of a critical upstream bug.

Biosecurity
DNA Synthesis
Verifiable Credentials
Compliance
Metagenomic Surveillance
In Progress
View Paper

Emergent Misalignment from Political Content Fine-Tuning at 7B Scale

Pavan Kumar Dubasi

BlueDot Impact AI Safety Fundamentals Sprint 2026

Sole-authored research, BlueDot Rapid Grant recipient ($250)

Political content fine-tuning triggers 3x stronger emergent misalignment than insecure code at 7B scale. Cross-architecture replication on Qwen 2.5 7B and Llama 3.1 8B, validated by a 4-judge LLM panel and human evaluation with Krippendorff alpha 0.90.

AI Safety
Emergent Misalignment
Fine-tuning
LLM Evaluation
QLoRA
In Progress
View Paper

A Study on Facial Feature Extraction and Emotional Analysis Using ML

Adhyaru, K., Ishan, Harshitha, D., Mundlur, R.P., Dubali Pavan Kumar, et al.

International Journal of All Research Education and Scientific Methods (IJARESM)

ISSN: 2455-6211, Vol. 11, Issue 7, July 2023, pp. 2455-2467

This research presents a comprehensive approach to facial feature extraction and emotion recognition using machine learning techniques, achieving high accuracy in real-world scenarios.

Facial Recognition
Emotion Analysis
Machine Learning
Computer Vision
Published
View Paper

Federated Learning for Privacy Preserving Collaboration in Healthcare

Pavan Kumar Dubasi, et al.

2025 International Conference on Communication, Computer Sciences and Engineering (IC3SE-2025)

19-21 March 2025, Amity University Uttar Pradesh, Greater Noida, India

This paper presents a novel federated learning approach for healthcare data collaboration that preserves patient privacy while enabling effective model training across multiple healthcare institutions.

Federated Learning
Healthcare
Privacy
Machine Learning
Accepted

Identity Delegation Attacks in ControlArena Evaluation Environments

Pavan Kumar Dubasi

AI Control Hackathon 2026 (Apart Research + Redwood Research)

48-hour hackathon, Apr 2026

Develops a ControlArena evaluation environment for identity delegation attacks. Discovers that format mismatches between attacker and monitor silently kill all monitoring scores. Behavioral monitor catches 50% of attacks at threshold 0.5.

AI Control
Identity Delegation
ControlArena
Security Evaluation
Completed
View Paper

NAO Metagenomic Surveillance Pipeline Audit

Pavan Kumar Dubasi

BlueDot Impact Biosecurity Programme 2026

Technical Report, Feb 2026

Independent deployment and comprehensive audit of the Nucleic Acid Observatory metagenomic surveillance pipeline. Discovered a critical upstream bug affecting sequencing data processing and documented full reproducibility steps for the bioinformatics workflow.

Biosecurity
Metagenomics
Surveillance Pipeline
Bioinformatics
NAO
Completed

Patents

2

AI-Enabled Robotic Device for Business Automation

Kousar, H., Pavan Kumar Dubasi, Mohammed, S.A., et al.

UK Design Registration No. 6335588, Grant Date: 10 January 2024

An innovative robotic system that leverages AI for automating complex business processes, improving efficiency and reducing operational costs.

AI
Robotics
Automation
Business Process

Emotion Recognition and Analysis System for Mental Health Assessment

Navaneethakrishan, S.R., Chinta, S., Pavan Kumar Dubasi, et al.

The Patent Office Journal, 29/09/2023, No. 39/2023, pp. 64318-64802

A novel system for real-time emotion recognition and analysis, specifically designed for mental health monitoring and assessment.

Emotion Recognition
Mental Health
AI
Healthcare
Published
View Patent

Professional Memberships

6

Association for the Advancement of Artificial Intelligence (AAAI)

ID: 650314

2025 - Present

Association for Computing Machinery (ACM)

ID: 8485149

Aug 2024 - Present

American Mathematical Society (AMS)

ID: PVDBXA

Aug 2024 - Present

International Association of Engineers (IAENG)

ID: 383517

Aug 2024 - Present

Institute of Electrical and Electronics Engineers (IEEE)

ID: 98915358

Jan 2024 - Present

International Society for Computational Biology (ISCB)

ID: 45957

2025 - Present

Academic Profiles

Find my full publication record on these platforms.