I work on how large models learn and reason.
I’m Tianle, a Member of Technical Staff at Thinking Machines Lab, where I lead the Auto Research team. I work on automating ML research, reinforcement learning, and post-training.
01
Selected work
Recent contributions
02
About
A short history
I’m interested in closing the loop between how models are trained and how ML research itself gets done.
At Thinking Machines Lab, I was the model captain for Inkling and started the reasoning team. I now lead Auto Research, working to automate ML research.
Previously, I worked on the Grok 4 family at xAI. Before that, I studied EECS at UC Berkeley, where I was advised by Ion Stoica and helped build LMArena.
03
On my mind
Open questions
-
A
Can we get more than one bit of information per rollout in reinforcement learning?
-
B
Can we skip mid-training? Can a model learn on the fly?
04
Writing
Notes from the field
A Hitchhiker’s Guide to Undergrad Research
What I wish I had read before starting research—from finding mentors to navigating PhD applications.
Elsewhere