| 3:30 PM | 🎤 Speaker Karthigayan Devan: Prompts as Observable, Policy-Enforcing Control Planes for AI Behavior | Large Language Models are increasingly embedded into production systems, yet most teams still treat prompts as informal chat instructions rather than operational artifacts. This creates a familiar SRE failure mode: powerful systems with no guardrails, no observability, and no policy enforcement.
In this talk, we reframe prompts as control planes for AI behavior, borrowing proven ideas from Site Reliability Engineering and Platform Engineering. You’ll learn how to design prompts that don’t just generate answers, but enforce architectural policy, surface assumptions, expose uncertainty, and constrain blast radius, the same principles we apply to production infrastructure.
We’ll explore how to instrument prompts for observability (assumptions, missing inputs, confidence levels), encode reliability, security, and cost policies directly into prompt contracts, and make AI outputs auditable, reproducible, and debuggable. Through real-world examples, architecture reviews, incident analysis, and internal developer platforms, you’ll see how policy-driven, observable prompting dramatically improves trust and reliability in AI-assisted systems.
This session is for engineers who believe AI should be treated like any other production dependency: governed, observable, and resilient by design. |
| 4:30 PM | 🎤 Speaker Sanket Bisne: Building the Real world Applications using Gemini3 and Google AI studio | With the rapid evolution of generative AI, Gemini 3 and Google AI Studio are redefining how developers build intelligent, production-ready applications.
In this session, we will explore how to design and build real-world applications using Gemini 3, leveraging Google AI Studio for rapid prototyping, prompt engineering, and model orchestration. The talk will move beyond theory and focus on practical use cases, including conversational agents, content generation, workflow automation, and AI-powered developer tools.
Attendees will gain hands-on insights into:
How Gemini 3 works and where it fits in modern application architectures
Using Google AI Studio to experiment, fine-tune prompts, and test models
Integrating Gemini APIs into real-world applications
Best practices for performance, security, and responsible AI usage
By the end of the session, participants will be equipped to start building scalable, AI-powered applications using Google’s latest AI ecosystem. |
