Suraj Parmar

I work in applied AI at Maisa, mostly fine-tuning small models and the evaluation frameworks around them.

Before that, two years of medical AI research at SpassMed, with teams in South Korea: air-gapped RAG for hospitals, and models to catch sepsis earlier from ICU data.

Further back, a data scientist at RealAdvisor working on real estate valuation, and a Master's at the University of Windsor spent on deep learning over Numerai's encrypted market data.

Machine learning has been the constant. Everything else I read tends to follow whatever problem is in front of me at the time, which is how I ended up writing about molecules.

Find me on X, GitHub, LinkedIn, or Hugging Face. I write sometimes on Substack and Medium, and post the occasional video on YouTube.