Research & Publications

Work at the intersection of peer-review transparency, argumentation theory, and applied GenAI systems.

Published

Dynamic Optimization of Peer Review Length Using Information Density Analysis

Scientometrics (Springer)

Introduces an information-density framework that balances relevance, argumentative strength, and cognitive load to optimise peer-review length.

Under Review

AntiBug: Runtime-Aware Multimodal Agentic Program Repair for Production Applications

Automated Software Engineering (Special Issue)

Proposes a novel automated software patching framework using multimodal runtime telemetry (stack traces, screenshots) and specialized agent orchestration for hypothesis-driven repair.

Under Review

FRAC-MAS: Fracture Radiograph Analysis using Conformal Multi-Agent System

Biomedical Signal Processing and Control (Special Issue)

Introduces a conformal multi-agent fracture radiograph pipeline integrating robust visual models with specialized LLMs, ensuring statistically grounded diagnoses for clinical workflow integration.

Under Review

BiasChain: A Multi-Agent LLM Framework for Justified Peer Review Bias Detection

Journal of Information Science (Sage Publications)

Automates bias detection in peer reviews by orchestrating specialized agents for sentiment coherence, internal consistency, and inter-review alignment.

Under Review

DigniFy: Multi-Modal Multilingual Online Hate Speech Detection

IEEE CONECCT

Presents a theoretical framework and pipeline for multilingual, multimodal safety systems with interactive LLM orchestration to detect online hate speech.

Whitepaper

Evaluating Large Language Models for Automated Requirement Generation in IT Projects

LinkedIn (Nvelop Technologies Oy)

A comprehensive benchmark comparing DeepSeek R1, GPT-4o, Gemini 2.0, and LLaMA 3.2 on BLEU, Levenshtein, and Jaccard metrics for requirement engineering.

Achievements & Recognition

GATE DS&AI Rank 499

Ranked 499 nationwide among 60,000+ candidates in GATE Data Science & AI, demonstrating strong theoretical foundations.

Amazon ML Challenge Rank 292

Placed 292 out of 70,000+ participants in Amazon ML Challenge 2024, showcasing applied machine-learning expertise.

CodeBounty 2024 Winner

Won the flagship competitive programming contest at DJSCE, highlighting algorithmic design strengths.