# Dhruvi Paprunia > Public portfolio, research, CV, and plain-language AI paper breakdowns. This index points to public source pages and Markdown copies. Current HTML is authoritative. Proposed tests in articles are not reported experimental results. ## Portfolio and writing - [Hi :) I'm Dhruvi I teach small models to do big things](https://dhruvipaprunia.com/index.md): Applied AI researcher. RL for small language models, and AI that ships. - [Hi :) this is where I overshare about AI.](https://dhruvipaprunia.com/blog/index.md): Notes on RL, small language models and AI that works. - [30 days, 30 papers.](https://dhruvipaprunia.com/series/index.md): One research paper a day, explained simply. - [A million tokens can still lose the right document](https://dhruvipaprunia.com/blog/a-million-tokens-can-still-lose-the-right-document/index.md): Why seeing every document is not the same as finding the right one. - [Your model found the right answer. It might never learn to find it again.](https://dhruvipaprunia.com/blog/a-rare-correct-answer-can-still-get-lost-in-training/index.md): Why GRPO, training that compares rewards across attempts, can help one try more than many. - [Can one reward teach an assistant everything?](https://dhruvipaprunia.com/blog/can-one-reward-teach-an-assistant-everything/index.md): A reading of MAHALO, where accuracy, human values and tutoring pull an assistant in different directions. - [A coin flip made this model better at math. Or did it?](https://dhruvipaprunia.com/blog/coin-flip-math/index.md): Notes on - [The polite sentence that broke my model](https://dhruvipaprunia.com/blog/grpo-tool-use/index.md): How we taught small language models to call tools with GRPO, and why the strictest rule worked best - [Never Give Up: why easy problems eat the RL budget](https://dhruvipaprunia.com/blog/never-give-up/index.md): A deep dive into the Matthew Effect, and how adaptive sampling gives hard problems another chance. - [The one percent we stopped looking at](https://dhruvipaprunia.com/blog/one-percent-we-stopped-looking-at/index.md): Muon, Never Give Up, and the cases an average can hide. - [The agent did everything right. The website still said no.](https://dhruvipaprunia.com/blog/the-agent-did-everything-right-the-website-still-said-no/index.md): What a verifier can learn when the assistant took the right steps but the website failed. - [The model weights are open. Can we see how the model learned?](https://dhruvipaprunia.com/blog/the-model-weights-are-open-can-we-see-how-the-model-learned/index.md): Why open model training matters, and how to test the steps that shape an assistant. ## Other public resources - [CV](https://dhruvipaprunia.com/5-Dhruvi-Paprunia-CV.pdf): public curriculum vitae. - [RSS](https://dhruvipaprunia.com/feed.xml): writing feed. - [Sitemap](https://dhruvipaprunia.com/sitemap.xml): public HTML page list. ## Read-only use Public crawling, including search and model-training crawlers, is allowed in robots.txt. No public agent action API is provided. Opening a contact link or interactive guestbook does not grant permission to submit a message.