Scalable Authorization Platform & Distributed Intelligence in AI-Powered Systems

Hi Everyone,
We’re back with the second edition of Expert Talks for the year—and this one is extra special. We’re excited to host guest speakers from Freshworks for an engaging, in-person session with the community.
Please join us on 28 February 2026, from 10:30 AM to 1:00 PM @ Four Points By Sheraton Chennai OMR - Vaigai Hall.

Agenda :
1. Welcome & Intros (10 mins)
2. Talk 1: Building a State-of-the-Art Authorization Platform with Graph Database for Scale (45 mins)
3. Tea Break (15 mins)
4. Talk 2: Distributed Intelligence in AI-Powered Software Systems (45 mins)
5. Networking

Talk 1: Building a State-of-the-Art Authorization Platform with Graph Database for Scale

At Freshworks, which powers over 73,000 businesses worldwide, authorization is on the critical path of every user interaction. As our product portfolio grew, authorization logic evolved independently across teams—leading to inconsistent behavior, scalability limitations, and increasing maintenance costs.
In this talk, I’ll walk through how we designed and built Platform AuthZ, a centralized authorization platform capable of serving millions of fine-grained authorization checks per day in under 50 ms. We’ll explore why traditional approaches fell short at SaaS scale, how graph-based modeling using Amazon Neptune enabled expressive and dynamic authorization decisions, and the architectural choices that helped us balance correctness, performance, and extensibility.

Talk 2: Distributed Intelligence in AI-Powered Software Systems (How AI Changes Where Decisions and Reasoning Live)

Software systems used to behave in ways we could explain: a decision was encoded in code, a rule, or a design choice. When something went wrong, we knew where to look.
AI has quietly changed that.
Today, many AI-powered systems feel harder to reason about. Everything appears correct, yet outcomes can still surprise us. What changed in the structure of our systems?
This talk introduces a systems-thinking lens for understanding how AI redistributes decision-making in modern software systems. Rather than focusing on tools or prompt techniques, it explores how design choices shape autonomy, reliability, and feedback in systems that incorporate LLMs.
The session is intended for engineers working with AI-assisted software development who want a clearer mental model for reasoning about system behavior, constraints, and long-term maintainability.