Vadivel Abishethvarman

I am a researcher and lecturer specialising in Natural Language Processing (NLP), Large Language Models (LLMs), and AI safety, based in Sri Lanka. I hold a BSc (Hons) in Computing and Information Systems (First Class) from Sabaragamuwa University of Sri Lanka, where I was awarded the Gold Medal for Best Research in Computing at the 25th General Convocation (2024).

My research spans knowledge distillation, LLM safety and alignment, fake news detection, sentiment analysis, and medical imaging — with over 50 peer-reviewed publications in journals and conferences including Springer Nature, IEEE, MDPI, and ACL venues. I collaborate with researchers across Sri Lanka, Australia, Japan, and India.

Research Interests

• Large Language Models — safety, alignment, knowledge integrity, guardrails
• Knowledge Distillation — efficient NLP, compact models, task-specific distillation
• Fake News & Misinformation — detection, categorisation, user profiling
• Medical Imaging — deep learning for fracture detection and classification
• Sentiment & Social Media Analysis — public perception, topic modelling

News & Highlights

• 🏆 May 2026 — Paper published in Machine Learning (Springer Nature): Survey on LLM Safety: Attacks, Defenses, Alignment, Metrics, and Guardrails
• 📄 Dec 2025 — Paper published in Information (MDPI): Knowledge Integrity in Large Language Models: A State-of-The-Art Review
• 🎤 2025 — Best Presenter Award, 2nd International Conference on Technological Research & Innovation (ICTRI 2025)
• 🖼️ 2025 — Best Poster Award, 13th Annual Research Session (ARS 2025), SUSL
• 📄 2024 — Best Abstract Award, 1st Computing Undergraduate Research Symposium (ComURS 2024)
• 🥇 Sep 2024 — Gold Medal for Best Research in Computing, 25th General Convocation, SUSL

Affiliations

Selected Publications

Survey on LLM Safety: Attacks, Defenses, Alignment, Metrics, and Guardrails Machine Learning, Springer Nature, 2026 — doi.org/10.1007/s10994-026-07060-8

Knowledge Integrity in Large Language Models: A State-of-The-Art Review Information, MDPI, 2025 — doi.org/10.3390/info16121076

XGUARD: A Graded Benchmark for Evaluating Safety Failures of LLMs on Extremist Content Submitted to ACL, 2025 — arxiv.org/abs/2506.00973

Unmasking Fake News with Emotional Cues ICARC 2024, IEEE — doi.org/10.1109/ICARC61713.2024.10499790

→ Full list at abishethvarman.github.io/publication/

Awards & Recognition

→ View certificates at abishethvarman.github.io/awards/

“Research is not just about publishing — it is about pushing the boundaries of what is known and making that knowledge accessible.”