Frank Mueller

About Frank Mueller

Most teams building AI-powered products treat collaboration as an afterthought. Bolt on a co-editing layer, handle conflicts later. My team builds the infrastructure that makes real-time collaboration between humans and AI agents actually work at scale.I lead engineering for Fluid Framework, Azure Fluid Relay, and new collaborative AI experiences within Microsoft 365 Copilot. The infrastructure my teams own powers real-time collaboration in Word, Copilot Pages, and Loop at 1B+ sessions per month. The problem we\'re solving now: when humans and AI agents co-author simultaneously, you need merge semantics, attribution, and permissions designed for participants that write at machine speed, operate across multiple sections at once, and have no social norms about what they edit.Before this, I spent nearly seven years at Amazon, where I built a 35-engineer team that launched Amazon GameSparks as a new AWS service from zero to GA, and earlier built the engineering org behind Amazon\'s last-mile delivery expansion including Flex. Before Amazon, I led engineering for search and content distribution at Yahoo, and started my career at Microsoft building OneNote and InfoPath, an early education in the collaboration space I\'m now back in.What I\'m most focused on right now: defining how AI agents participate in collaborative sessions as first-class participants, not tools invoked from a sidebar. And a newer question my team is tackling: how do multiple users collaboratively edit AI-generated interactive experiences in real time? That intersection of generative AI and real-time collaboration is where I spend most of my time.Always open to conversations with people working on real-time collaboration, multi-agent systems, or human-AI co-authoring.

Experience

1. Group Engineering Manager

   Microsoft

   Mar 2022 — Present · Seattle, WA, US

   I lead the engineering teams behind Fluid Framework, Azure Fluid Relay, and new collaborative AI experiences in Microsoft 365 Copilot.Fluid Framework is Microsoft\'s open-source, real-time collaboration runtime. My teams own the distributed infrastructure that keeps multiple participants in sync across M365, handling consistency, conflict resolution, and session management at scale. As AI agents become active participants in collaborative sessions alongside humans, that infrastructure is taking on an entirely new set of challenges: higher write throughput, new attribution and permissioning models, and interaction patterns that didn\'t exist two years ago.On the Copilot side, my teams are building the experiences where these capabilities meet users. The work spans from the collaboration layer up through the UX, in close partnership with research, design, and product.

Education

• Purdue University

  Bachelor of Science, Computer Sciences, Mathematics

• Purdue University

  Master of Science, Computer Sciences