RSI Research Mixer
RSI Research Mixer
A curated gathering for researchers, engineers, founders, and technical builders exploring recursive self-improvement and self-learning AI systems.
AI agents are becoming more capable, but the harder question is how they learn from experience, retain useful knowledge, evaluate their own work, and improve safely over time.
The RSI Research Mixer brings together people working across research, infrastructure, developer tools, reinforcement learning, evaluations, memory, and agentic systems for an evening of practical conversations and shared learning.
This is not a pitch night. It is a research-led gathering for people actively building, experimenting, or trying to understand what self-improving AI systems could look like in real-world environments.
What we’ll explore
Recursive self-improvement and self-learning agents
Memory, feedback loops, and continual learning
RL environments and world models
Evaluations, verification, and agent reliability
Developer infrastructure for long-running agents
Moving research ideas into practical systems
Event format
Spotlight Talks
Short sessions from researchers and builders sharing what they are working on, what they have learned, and where they are still stuck.
Research Discussion
An open conversation around the technical opportunities, limitations, and safety considerations of self-improving AI systems.
Open Networking
Meet researchers, founders, engineers, independent builders, and people exploring the next generation of agentic systems.
Who should attend?
AI and ML researchers
Engineers building agentic systems
Founders and startup operators
Developers working on AI infrastructure and tooling
People exploring RL, evaluations, memory, or world models
Independent builders and open-source contributors
📍 Paytm Office, Bellandur
📅 October 24, 2026
Hosted by AgentR
Ecosystem Partner: Paytm AI
Interested in presenting your research, sharing an experiment, or joining the discussion? Contact the host and tell us what you would like to speak about.