Orion Weller

I am a researcher at Google DeepMind working on Gemini, where I’ve contributed to the post-training of the 3.5, 3.6, 3.7, and 3.8 series of models. My current research focuses on long-context capabilities, agentic behavior, and model collaboration for upcoming Gemini models.

Previously, I completed my PhD at the Center for Language and Speech Processing at Johns Hopkins University, advised by Benjamin Van Durme and Dawn Lawrie and graciously supported by an NSF Graduate Research Fellowship. My PhD research focused on the intersection of Information Retrieval (IR) and Large Language Models (LLMs) with a focus on instruction-based and reasoning-aware retrieval (Promptriever, Rank1, LIMIT), encoder model pre-training (ModernBERT, mmBERT), and model evaluation (FollowIR, Dated Data).

I was fortunate to intern with great mentors at Meta FAIR, Google DeepMind, AI2, Samaya, and Apple. Earlier, I received my B.S. in CS and statistics from BYU, advised by Kevin Seppi and Quinn Snell.

If you’re interested in getting in contact with me, please email me at {last_name}{first_name}@gmail.com.