Research & Publications

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

Accepted

Dynamic Optimization of Peer Review Length Using Information Density Analysis

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

Under Review

MetaSearch: Search-Augmented LLM with Reasoning for Consensus Resolution in Peer Review

IIT Patna Collaboration

Combines disagreement detection, fact-grounded synthesis, and argument evaluation to automate meta-review decisions.

Under Review

From Black Box to Glass House: Argumentation Theory as a Bridge Between LLM Opacity and Peer Review Transparency

IIT Patna Collaboration

Aligns argumentation theory with LLM signals to improve review interpretability, achieving strong agreement with human annotations.

Under Review

Unmasking Bias: LLM-Driven AI Agents in Academic Peer Review

IIT Patna Collaboration

Automates bias detection in peer reviews by analysing sentiment, coherence, and consistency, supporting organisers with actionable insights.

Whitepaper

LLM-Based Requirement Generation: Evaluation Pipeline and Benchmarking

Nvelop Technologies Oy

A 70-page enterprise benchmark comparing DeepSeek R1, GPT-4o, Gemini 2.0, and LLaMA 3.2 on BLEU, Levenshtein, and Jaccard metrics for requirement engineering.

Under Review

DigniFy: Multimodal Hate Speech Detection in Digital Platforms

College Research Collaboration

Explores multilingual, multimodal safety systems with real-world platform evaluations and policy-aligned mitigation insights.

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.

Rajasthan Police Hackathon Finalist

Led ML for the finalist solution tackling law-enforcement challenges with data-driven citizen feedback analysis.

sktime Mentoring Program

Selected as a mentee under Dr. Franz Kiraly for the sktime mentoring program focused on time-series ML research.