Alexander Peysakhovich

Currently I am a Researcher in Residence (Chief Matrix Multiplier) at Sutter Hill Ventures. I help our portfolio companies prepare their AI strategies, help them build out production systems, and prototype new ideas.
Before this I was a Research Scientist at Meta FAIR working on machine learning. Before that I was the technical lead of the Facebook "Feed Science" team where we built, rolled out, and analyzed various machine learning improvements to News Feed. Before studying AI, I worked on human cognition. I got my PhD in Behavioral Economics from Harvard.
You can reach me at alex (dot) peys (at) gmail.com
Research Interests
I have worked on many different machine learning and related topics. Below I list some current areas of interest with links to representative publications. For a full list of publications see Google scholar.
- Multimodal models - using text and image as both inputs and outputs of large models [1]
- Embeddings and retrieval - graph-based learning, retrieval augmented generation, and non-Euclidean embeddings [1], [2]
- Multiagent environments - building artificial agents that know how to coordinate, communicate, and cooperate with others [1], [2], [3]
- Machine Learning and Causality - applications to complicated environments like advertising markets, large scale A/B testing, or medicine [1], [2]