Alvaro Velasquez
University of Colorado Boulder, CO·Computer Science
Artificial intelligence (AI) systems are increasingly built as teams of agents that split up work, share information, use tools, and make decisions over many steps. Most such multi-agent systems still depend on a central controller to hand out tasks.
- Fields
Computer Science
Engineering
Mathematics & Statistics
- Topics
multi-agent systems
decentralized AI
self-organizing agents
- Grants
- 1 known grant, through Sep 2029
What they work on
- Develop decentralized self-organization methods in which AI agents decompose tasks among peers, assign and enhance roles, and self-refine during execution, using information-theoretic tests to judge whether emergent behavior is useful.
- Build a serverless, peer-to-peer training framework in which multi-agent AI systems recover from failures, reorganize dynamically, and share knowledge without harmful interference.
- Treat agent roles, such as system prompts, as latent variables, using simulation-based inference, continual consensus and multi-agent direct preference optimization to discover role configurations that produce desired collective outcomes.
More about the research
That creates bottlenecks as systems grow and single points of failure. It also makes collaboration hard when agents have different skills or cannot share everything.
Why it matters
Decentralized AI agent teams could be more scalable and resilient than centrally controlled ones because no single failure can bring them down. The grant names possible uses in healthcare, transportation, smart grids, smart cities, social simulation and scientific discovery. Better collective AI could support these fields while helping avoid harmful group behavior.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Craig Newgard
Oregon Health & Science University, OR·Computer Science
Develop a reinforcement learning framework in which two agents select features and a third selects an operation, with a shared value network, to generate new features in a traceable way.
Navid Azizan Ruhi
Massachusetts Institute of Technology, MA·Engineering·Early-career
Develop Hard-Constrained Neural Networks (HardNet), which use a differentiable projection layer to guarantee input-dependent constraints, and embed them in reinforcement learning to produce certifiably safe policies.
Lingming Zhang
University of Illinois Urbana-Champaign, IL·Computer Science
Train LLMs jointly to generate synthetic software issues and to resolve them, using a self-play loop in which the model learns to create the issues that most improve its own bug-fixing ability.
The school
University of Colorado Boulder: 36 labs on PI Pages, $718M a year on research, 447 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.