Building a Gi-ralph Loop for Software Development

Hosted by Amir Feizpour and Joel Fernandes

Fri, May 15, 2026

4:00 PM UTC (45 minutes)

Virtual (Zoom)

Free to join

Invite your network

Go deeper with a course

Agentic Buildcamp - Build AI Agents Using AI Agents (Claude Code)
Amir Feizpour, PhD
View syllabus

What you'll learn

What a Ralph Loop Is (and Isn’t)

A clear mental model of iterative agent loops and how they differ from simpler workflows.

Designing Stable Long-Running Loops

Key architectural patterns for building loops that don’t degrade over time.

Introducing the “Giralph” Loop

A new loop design (greatly improved ralph) aimed at improving robustness and outcome quality.

Failure Modes in Practice

How modern models can still destabilize loops—and how to detect and mitigate this.

When Not to Use Ralph Loops

Identifying cases where simpler approaches outperform complex iterative systems.

Why this topic matters

As agentic systems move toward longer-running, self-improving workflows, understanding loop design becomes critical. Ralph loops represent one of the clearest patterns for extending model capability—but also expose the limits of current systems. This session offers a practical perspective on how to use them effectively without overengineering fragile solutions.

You'll learn from

Amir Feizpour

Founder @ Aggregate Intellect

Amir Feizpour is the founder, CEO, and Chief Scientist at Aggregate Intellect building a generative business brain for service and science based companies. Amir has built and grown a global community of 5000+ AI practitioners and researchers gathered around topics in AI research, engineering, product development, and responsibility. Prior to this, Amir was an NLP Product Lead at Royal Bank of Canada. Amir held a research position at University of Oxford conducting experiments on quantum computing resulting in high profile publications and patents. Amir holds a PhD in Physics from University of Toronto. Amir also serves the AI ecosystem as an advisor at MaRS Discovery District, works with several startups as fractional chief AI officer, and engages with a wide range of community audiences (business executives to hands-on developers) through training and educational programs. Amir leads Aggregate Intellect’s R&D via several academic collaborations.

Joel Fernandes

Engineering Lead @ Nvidia

Joel Fernandes is a systems software engineer with over 16 years of experience, currently working at NVIDIA. His expertise spans core Linux kernel subsystems including scheduling, memory management, synchronization, and tracing. He works on the Nova Rust GPU driver and develops kernel-level software to run large language models efficiently on NVIDIA hardware, bringing a deep systems perspective to AI infrastructure.

See all products from aggregate intellect

Sign up to join this lesson

By continuing, you agree to Maven's Terms and Privacy Policy.