# Zexi "Jesse" Zhang

> Robotics researcher and ML engineer focused on humanoid loco-manipulation and reinforcement learning.

## About

Zexi "Jesse" Zhang is a researcher at Imperial College London and the Swirl Lab, working on enabling humanoid robots to perceive, reason, and act in complex environments. His research combines vision-language models with reinforcement learning for robotic control. He also maintains Gemini Voyager, an open-source browser extension for Google Gemini with 12,000+ GitHub stars, and writes a technical blog covering LLMs, RL, diffusion models, and GPU programming.

## Author

- Name: Zexi "Jesse" Zhang (Nagi-ovo)
- Affiliation: Imperial College London / Swirl Lab
- Site: https://www.nagi.fun
- GitHub: https://github.com/Nagi-ovo
- Google Scholar: https://scholar.google.com/citations?hl=en&user=jktWnL8AAAAJ

## Research

- Humanoid loco-manipulation and reinforcement learning
- Vision-language navigation (TagaVLM)
- Intelligent agent perception, reasoning, and action in complex environments

## Publications

- TagaVLM: Topology-Aware Global Action Reasoning for Vision-Language Navigation (ICRA 2025)

## Projects

- [Gemini Voyager](https://github.com/Nagi-ovo/gemini-voyager) — All-in-one enhancement suite for Google Gemini (17,000+ stars)

## Related Sites

- [blog.nagi.fun](https://blog.nagi.fun) — Technical blog on ML, LLMs, reinforcement learning, and deep learning
- [blog.nagi.fun/llms.txt](https://blog.nagi.fun/llms.txt) — Blog knowledge file for AI agents
- [voyager.nagi.fun](https://voyager.nagi.fun) — Gemini Voyager project site
- [GitHub](https://github.com/Nagi-ovo) — Open-source projects and code