I'm Vidhya MN, a computer science engineer who builds generative AI and NLP applications — from diagnostic imaging models to legal-query assistants — and ships them end to end, not just in notebooks.
I graduated from Sambhram Institute of Technology with a degree in Computer Science, and spent the last few months of that time in an intensive AI internship building generative and agentic applications that had to work outside of a Jupyter notebook — provisioned, tested, and handed off.
My final-year project used a GAN-based Pix2Pix model to generate virtual pathology images, aimed at cutting diagnostic turnaround time for breast cancer testing. That project set the pattern for how I like to work: pick a real bottleneck, and build a model that actually addresses it, not just one that scores well on a benchmark.
Outside of AI, I've spent three years in the National Cadet Corps and represented my college in Kabaddi at the Taluk level — both taught me a version of discipline that shows up in how I debug: patiently, and without skipping steps.
Immunohistochemistry (IHC) testing is accurate but slow and expensive. This project explored whether a generative model could produce reliable virtual IHC images directly from routine H&E slides — cutting cost and turnaround time without sacrificing diagnostic signal.
Legal information in India is often locked behind dense language and paywalled consultation. Just Nest is an NLP-based assistant that answers legal queries and surfaces relevant case information in plain language, in real time.
Most budgeting apps show you what you spent. Cash Clue tries to show you what you're about to spend — using a machine learning model over transaction history to surface patterns and forecast where money is heading.