vikhram-labs/NariRaksha-3B
Text Generation • Updated • 1
AI & ML interests
Multilingual AI, Efficient Foundation Models, Low-Resource NLP, Reasoning Systems, RAG, Biomedical AI, Sovereign AI Infrastructure
Recent Activity
Organization Card
Vikhram Labs
Multilingual AI · Efficient Foundation Models · Low-Resource NLP · Reasoning Systems · RAG · Biomedical AI · Sovereign AI Infrastructure
Website · GitHub · Google Scholar · LinkedIn · ORCID
About
Vikhram Labs is the research organization of Vikhram S, an AI researcher working at the intersection of multimodal intelligence, explainable machine learning, and public-interest technology. The lab builds AI systems for high-stakes, real-world domains — medicine, law, and public safety — with an emphasis on transparency, reliability, and reasoning under uncertainty.
Vikhram S earned his B.E. in Electronics and Communication Engineering from Saveetha Engineering College, India (2026), where he now advises the Tech Society on machine learning research and mentors undergraduate researchers.
Research Focus
- Vision–Language Models & Explainable AI — interpretable multimodal architectures for clinical decision support
- Multilingual & Low-Resource NLP — models and benchmarks for underrepresented languages, including Indic languages
- Reasoning Systems & RAG — retrieval-augmented, confidence-aware inference pipelines
- Biomedical AI — dermatological diagnosis, radiology report synthesis and interpretation
- Sovereign AI Infrastructure & Legal NLP — constitutional and civic-technology AI systems
- AI for Public Safety & Gender Empowerment — applied AI for social impact
Featured Work
| Project | Description |
|---|---|
| NariRaksha-3B | Model powering gender-responsive AI research for women's safety |
| NariRaksha-100K | Dataset supporting the NariRaksha initiative |
| IndicConBench | Benchmark dataset for Indic-language conversational AI |
| IndianConstitution (PyPI) | Open-source constitutional/legal NLP package — 33,000+ downloads |
Selected Publications & Recognition
- NariRaksha: Gender-Responsive AI for Women's Safety — India AI Impact Summit 2026 Compendium on AI and Gender Empowerment, jointly published by India's Ministry of Electronics and Information Technology (MeitY) and UN Women. Selected as 1 of 23 case studies from 235+ submissions across 50+ countries. Read →
- Explainable SLM-Guided Vision–Language Model for Multi-Class Skin Lesion Recognition — Springer Lecture Notes in Networks and Systems, ICICC 2026. Ranked in the top 15% of papers reviewed; indexed in Scopus, DBLP, zbMATH, EI Compendex, and SCImago. DOI →
- Explainable Vision-Language Generative Models for Radiology Report Image Synthesis and Interpretation — journal submission in progress
Full publication list on the personal site.
Talks & Engagement
Vikhram has led sessions at Saveetha Engineering College on AI Agents and Open Source Ecosystems, Quantum Machine Learning, Financial Machine Learning, Reproducible Deep Learning, and Cyber Law & Compliance for AI systems — see the full list on the Talks and Lectures page.
He has also engaged in research and policy discussions with leaders from NVIDIA, Microsoft Research Africa, Google Africa, the World Bank, UN Women, and the University of Pretoria — see Media.