Ahmed Medhat
Building coordination technology for humans and agents.
I'm researching coordination technology for networks of humans, models, and tools. The systems we inherited for coordinating were not built for the speed and interdependence of the world we now live in, and I am interested in what better ones could look like.
My career over the past fifteen years has been guided by one mission and one insight.
The mission: To bring coordination technology that can change how the world works for the better. Human coordination is our most consequential technology, yet the world we've built now exceeds the coordination systems we inherited.
The insight: Network topology is collective intelligence. In other words, in goal-oriented networks of actors (including entire societies), who each actor is connected to, how strong that connection is, and what flows across it shapes the collective output of the network, often more than the sum of the singular intelligences inside it.
This was a product of my life experience. Growing up in Cairo imprinted on me how cultures that once achieved greatness can decay when the institutional structures connecting their inhabitants collapse, locking them into stable but pathological patterns of connection.
Studying at Oxford during the Arab Spring matured my view of how online networks reshape who connects to whom and how they collectively act, and allowed me to merge perspectives from machine learning and sociology to quantify collective intelligence strictly from network topology, in what I believe was one of the earliest attempts of its kind.
Through years of research work at Meta, I was then exposed to the role media of interaction can play in re-shaping the collective output of networks of people, and in understanding the structural failure modes of large networks.
All of this has shaped my present focus on solving human-agent coordination by building topology foundation models, where the network structure topology itself is the thing being designed and learned.