Aaditya Vardhan Das

Hi, I'm Aaditya — physics enthusiast turned data scientist, which honestly isn't that big a jump when you think about it.

I did my BE in Computer Science from Thapar University, and somewhere along the way got completely hooked on the intersection of math, physics, and machine learning. I still do Kaggle competitions when I have the time, mostly because there's no better way to stress-test your intuitions.

Currently I'm a Data Scientist at ProcDNA, where most of my work sits at the intersection of clinical data and ML. The focus is deriving patient-level and HCP-level insights — things like longitudinal event prediction in rare and ultra-rare disease settings (when does a patient switch therapies, when do they drop off, what's the likely next diagnosis). There's also a decent amount of MLOps involved: maintaining batch ML systems on AWS and Databricks, building pipelines that actually survive production.

Right now I'm spending a lot of time on two things I find genuinely interesting: agentic AI system design, and what you could call inference engineering — getting large language models to run sensibly on constrained hardware. Both feel like problems where physics intuition actually comes in useful.