Haoling Pu

I'm a master's student at Carnegie Mellon's School of Computer Science and an ML systems engineer. I like the layer where the model meets the hardware and the product: kernels, serving, and the agents that sit on top.

These days I'm at CMU's Language Technologies Institute, working on speech translation that starts before the sentence ends — teaching a model to commit to words while the speaker is still talking. I'm also building Agent Anchor with BNY, a system that notices when AI agents quietly get worse, and contributing to SGLang-Omni, an open-source serving framework for speech models.

Last summer at Google I taught an agent to keep a team's knowledge base alive. Before CMU, computer science at the University of Michigan.

This site is where I keep the parts I find interesting. Email is the best way to reach me.