Model-Agnostic Systems: Building Your Moat When the LLM is a Commodity
As scaling laws commoditize intelligence, the real competitive advantage shifts from the LLM to the system architecture around it. This session presents a Model-Agnostic framework that treats foundation models as swappable utilities, prioritising upgrade velocity and cost optimisation over vendor loyalty. We’ll explore how to decouple product value from any single model provider so your stack stays resilient—regardless of which lab ships the next frontier leap. How do you design evaluations that make model swaps boring instead of terrifying? What does infrastructure look like when you can adopt a new frontier model the day it drops? And why are the best prompt engineers the ones who’ve automated themselves out of a job?
Shubh Chatterjee
Shubh is a technical leader with over a decade of experience building large-scale AI systems across financial services, lending, gaming, and fraud prevention. From real-time risk detection to conversational AI, marketing intelligence, and document processing, he has delivered solutions across the full spectrum of enterprise AI — where precision, security, and scale are non-negotiable. He turns complex machine learning theory into production-grade systems that solve real business problems, navigating the regulatory, operational, and technical constraints of the world’s most demanding industries.