Hi! I’m Yixin Dong, a Ph.D. student at Carnegie Mellon University, fortunate to be advised by Prof. Tianqi Chen. I hold a Bachelor of Science in Computer Science and Technology from the ACM Honors Class at Shanghai Jiao Tong University.

I am building efficient and verifiable AI agents. I regard it as a crucial pathway to AGI.

I am generally interested in ML systems and ML compilers. I am a dedicated advocate of open source and have built several adopted projects, including MLC-LLM and Apache TVM Unity, where I serve as an Apache TVM Committer. I had the privilege of spending a wonderful time at the University of Washington, working with Prof. Luis Ceze.

We have been driving the XGrammar project forward. It is a zero-overhead LLM constrained generation engine and has already been adopted by several well-known LLM frameworks and companies.

From 2025 to 2026, I worked at xAI (now SpaceXAI) and was the main contributor to xAI's structured generation effort. xAI adopted XGrammar for its structured generation library.

We are also currently pushing agent-driven kernel generation, including evaluation environment, sandbox infra, datasets, competition, agent workflow, compilers and DSLs, and post-training.

Feel free to explore my CV to learn more. I’m always open to collaborating with passionate, like-minded individuals.

I am open to advising CMU undergrads and master’s students with backgrounds or interests in ML systems and agents for coding. In particular, I am looking for 1-2 students interested in contributing to XGrammar. Experience in parser and system efficiency optimization would be a plus. I’m also happy to chat about research and PhD applications with students from underrepresented groups. Please reach out by email.

Beyond research, I’m a cinephile and a lover of contemporary dance. My favorite movies are Hiroshima mon amour and Il buono, il brutto, il cattivo. If you share similar interests, let’s grab a coffee and chat!