About

I'm Dhwanit Shah, a final-year B.Tech student in Computer Science and Business Systems at Bharati Vidyapeeth College of Engineering, Pune.

What this blog is

Every week I pick something new in AI (a model, a method, or a research paper) and test it myself with a small, honest experiment. Then I write up what I measured, what held up, and what didn't.

You'll find:

• Explainable AI: do explanation methods like Grad-CAM and LIME actually tell the truth?
• LLM evaluation: how reliable are the tools we use to judge models?
• Small and local models: what can a laptop-sized model really do?
• Indian-language AI: how well do models handle Hindi, Hinglish and Indian-accented English?
• Paper reproductions: rerunning published results to see if they hold.

Every post comes with its notebook or script, so you can rerun it yourself. Vendor claims and my own numbers are always kept separate.

About me

I build AI and backend systems and do research in explainable AI. I led a six-person team to a win at Smart India Hackathon 2024, worked as a Research & Engineering Intern at NTPC Limited, and I'm currently CIO at GreenFlag. On campus I lead Vigyanam, our science club, and serve as Treasurer of the ACM student chapter.

I'm open to roles and research collaborations in AI engineering and applied ML.

Stay in touch

• Newsletter: a monthly digest of what I tested (Substack link coming soon)
• LinkedIn: linkedin.com/in/dhwanit-shah-11b781291
• GitHub: github.com/Dhwanitisshah

---

Views are my own. Experiments are small-scale and for learning.