According to the web search results, Ulisse Mini is a 18 year old self-taught programmer who lives in South Carolin [6]. He is an autodidact (unschooled) and love to learn things quickly [2]. He currently works as a SERI MATS scholar at Stanford Existential Risks Initiative [1]. He has a website “uli.rocks” where he showcases his projects and experiments [2]. He is also active on GitHub [5], DEV Community, and LessWrong [3]. He likes to do “stupid stuff in the most complicated way possible” [5]. He is also a member of the AI Alignment Forum, a community blog devoted to technical AI alignment research [4]. Ulisse Mini also has a Twitter account under the handle @MiniUlisse where he describes himself as “born just the right time to save humanity” [6].

Some of his projects include:
- **understanding-analysis-solutions**: Solutions to Understanding Analysis by Stephen Abbott (second edition) [2][3][4].
- **montemac / activation_additions**: Algebraic value editing in pretrained language models [2][3][4].
- **ai**: Basic neuroevolution from scratch, playing openai gym games [2][3][4].
- **transformers**: From-scratch transformers with some experiments on latex, following karpathy's makemore [2][3][4]!
- **light**: Autodiff in pure lua. Think pytorch but a lot slower [2][3][4].
- **oth**: Obsidian To HTML, A template for building obsidian style notes to a static site [2][3][4].