I'm an AI researcher and M.S. AI student at Northeastern University, working at the intersection of vision-language models and biomedicine.

Right now, as a Research Assistant at Northeastern, I'm building an end-to-end vision-language model for veterinary fine-needle aspirate cytology — fine-tuning MedGemma 1.5 4B with QLoRA on a MedSigLIP encoder, deployed on Databricks with MLflow and Unity Catalog. Before that, my Research Capstone produced MorphoCLIP, a dual-encoder contrastive system aligning Cell Painting microscopy with natural-language perturbation descriptions across drugs, CRISPR knockouts, and ORF overexpressions — now a preprint on arXiv.

Before graduate school I spent a year at BigCircle (UPSAAS Technologies LLP) as an AI engineer, shipping a multi-agent Deep Research pipeline, dashboards, and React Native apps. The engineering background keeps the research honest — I think hard about pipelines, reproducibility, and what actually ships.

Outside the work above, my research interests span multimodal contrastive learning, parameter-efficient fine-tuning of large foundation models, attention mechanisms for medical image segmentation and classification, and the data engineering required to make terabyte-scale microscopy and pathology datasets tractable.