Head of go-to-market at Reactor, working on real-time interactive video and physical-AI infrastructure. He previously cofounded Nativ and worked in machine learning at Revolut, Heron Data and XYZ Reality.[5][3]
Start with The Next Medium: Why Real-Time Interactive Video Changes Everything →
Education
Master’s studies in Computer Science; withdrew
2018 – 2019
[4]Links
Website Conference presentation
X · Boudatw Head of Go-to-Market · Reactor
Career
Ahres studied computer science in the Netherlands, then pursued graduate study at EPFL. He left the master’s program to join Revolut full time. His subsequent work includes founding engineering at Heron Data, senior machine-learning engineering at XYZ Reality and Nativ, an AI-localization company. At EPFL he also assisted with teaching machine learning.[3]
Contributions
His Reactor talk distinguishes interactive video from generating a finished clip. User input changes a continuing stream, so the application maintains a bidirectional connection and returns frames as they are produced. Frame deadlines, inference latency and state consistency become interface constraints. The demonstrations explore navigable worlds, responsive characters and audience-driven applications, showing why the product must handle a live session rather than simply queue another batch-generation request.[1]
Tools, writing & talks
Writing, research & talks
Start here
A talk explaining live video input, streaming infrastructure and the application constraints of continuing generated worlds.
Start with the live-session architecture and the comparison with generating a finished clip. Basic API and streaming concepts help; the demonstrations provide a visual introduction before the infrastructure discussion.
Speaker[1]- The Next Medium: Why Real-Time Interactive Video Changes Everything — Ahmed Ahres, Reactoryoutube.com
The case that real-time generative video is a change of medium rather than a speedup, with the consequences that follow: streaming pixels instead of returning files, stateful sessions, and latency budgets that push GPUs close to users.
Catalog workOriginal source