ShubhamGajjar.
M.S. AI at Northeastern University.
Building practical AI for healthcare and real-world systems.
Researcher at the intersection
of vision & medicine.
I'm a published AI researcher with a year of production engineering experience. Most candidates have one or the other; I bring both to the same desk.
Currently building vision-language models for veterinary cytology.
As 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. The work covers the whole pipeline: multi-channel preprocessing, a 5-task VQA dataset curated from hierarchical pathologist annotations, and four cytologic grading systems encoded as chain-of-thought signals.
Latest across the desk.
Let's talk.
Open to research and engineering collaborations in medical AI, computer vision, and Large Language Models.
gajjar.shu@northeastern.edu ↗