Milan Varghese

Focused NLP & biomedical semantic-shift researcher: Milan's recent research centers on analyzing semantic change in biomedical language models (BioBERT/PubMedBERT) and applying NLP methods to social-computational problems such as internalized self-stigma on social media. milanvarghese+1

Methodological contributions in model initialization and LLM optimization: At Drexel (with Dr. Jake R. Williams) Milan developed a Bayesian Initialization algorithm for neural networks and is involved in optimizing large language models including custom attention mechanisms and ablation studies. milanvarghese+1

Industry-to-research engineer with end-to-end ML systems experience: Milan's background includes building backend infrastructure and enterprise ML solutions (Neuflo, EY), designing RAG/document-search pipelines on Azure, and prototyping generative- and deepfake-related systems—combining production engineering with research practice. weekday+1

Published computational social science researcher and active open-source/ML community participant: Milan has peer-reviewed/archival publications on sentiment analysis, a Google Scholar profile, and active profiles for code and model-sharing (GitHub, Hugging Face). google+2