Hello, I'm Julian!

I'm an undergrad at Stanford interested in large scale models of the human experience.

Most recently, I was a researcher at World Labs. Before that, I co-created Oasis, the first realtime world model with a playable demo. And prior to that, I optimized inference for the first era of large language models at Cohere and MosaicML.

Highlighted Work

	WorldGym: World Model as An Environment for Policy Evaluation Julian Quevedo, Ansh Kumar Sharma, Yixiang Sun, Varad Suryavanshi, Percy Liang, Sherry Yang [arxiv] [website] [code]
	Real-Time Frame Model World Labs [blog]
	Oasis: A Universe in a Transformer Decart & Quevedo, et al. [blog] [demo] [code] Press: TechCrunch
	Serving Quantized LLMs on NVIDIA H100 Tensor Core GPUs Nikhil Sardana, Julian Quevedo, Daya Khudia [blog]
	LLM inference performance engineering: Best practices Megha Agarwal, Asfandyar Qureshi, Nikhil Sardana, Linden Li, Julian Quevedo, Daya Khudia [blog]

github | twitter | google scholar | linkedin

I would love to hear from you: julianq@stanford.edu