Yinfei Yang

Also known as Yin-Fei Yang, 杨寅飞

Yinfei Yang is an AI researcher whose career links robot vision, semantic retrieval and multimodal foundation models. He studies Differential Evolution in Nanjing, researches machine-learning software and robot navigation, earns a Penn master's degree, and moves into engineering roles at Amazon and Redfin. At Google, he works on conversation-trained sentence representations, multilingual retrieval and ALIGN; at Apple, he coauthors MM1 and MM1.5 and contributes to Ctrl-R. As of October 2026, he is a co-founder and Chief Multimodal Architect at Elorian.

Now at: Elorian
Works: 169
Citations: 11,181
h-index: 30
Badges: Founder
Bridge
Wayfarer
Links: Website and Google Scholar

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

2007–2009

Differential evolution in Nanjing

From December 2007 to July 2009, he is a research assistant in the Evolutionary Computation Group at Nanjing University of Posts and Telecommunications. He compares Differential Evolution parameter settings and helps design ways to control them.
2010–2012

Brain imaging and robot navigation

From May to October 2010, he joins a collaboration on machine-learning software for epilepsy diagnosis using magnetoencephalography, with researchers at Nanjing University of Aeronautics and Astronautics and Tennessee Tech. From May 2011 to May 2012, he works as a research assistant to Temple professor Longin Jan Latecki on robot navigation, object tracking and human detection using a D-RGB camera.
2012–2017

Penn graduate study and industry roles

From 2012 to 2013, he studies for a master's degree at the University of Pennsylvania, with research in computer vision, robotics and natural language processing. From 2013 to 2015, he works as a software engineer at Amazon/Lab126. From 2015 to 2017, he is a data science engineer at Redfin, working on machine learning and computer vision.
2017–2021

Google sentence retrieval

From 2017 to 2021, he works as a research scientist at Google Research. In 2018, he writes about sentence representations learned from conversations and, in 2019, announces multilingual Universal Sentence Encoder modules for semantic retrieval.
2021

ALIGN and noisy text supervision

In 2021, he coauthors ALIGN, a vision-language model trained on noisy text supervision for visual and vision-language representation learning.
2022–2025

Apple multimodal foundation models

From 2022 to 2025, he is a research scientist and manager at Apple AI/ML, where he works on vision and language foundation models. In 2024, he coauthors MM1, a study of multimodal language-model pretraining. In 2025, he coauthors MM1.5; its specialized variants address video understanding and mobile UI understanding.
July 2026

Structured reasoning with Ctrl-R

In July 2026, he coauthors Ctrl-R, a framework that guides exploration of structured reasoning patterns in language and vision-language models. He does the work while at Apple.
As of October 2026

Co-founding Elorian

As of October 2026, he is a co-founder and Chief Multimodal Architect at Elorian.