Dive into the power of Gemma, Google's family of high-quality, lightweight open models designed for responsible AI innovation. This talk will explore how Gemma's performance make it an ideal backbone for building sophisticated agentic applications, from intelligent assistants to complex decision-making systems. We'll emphasize the advantages of leveraging these models for local development and deployment, discussing workflows, and best practices that enable developers to achieve exceptional quality, performance, and data privacy without relying solely on massive cloud resources.

In the year 2140, intergalactic travel is routine, and Earth’s safety hinges on a cutting-edge alien registration system. Agents J and S are on a mission to modernize this critical platform using AI-powered agents to ensure alien vessels and lifeforms are compatible with our planet. Join us to explore how ADK, Gemma, and serverless deployment (Cloud Run) power autonomous solutions for efficient alien screening. Whether you're passionate about AI, serverless technology, or sci-fi-inspired projects, this talk is your gateway to exploring how we protect Earth from intergalactic surprises. Don’t miss it!

For many Android developers, Jetpack Media3 is synonymous with ExoPlayer and video playback. But Media3 has grown into a powerful, modular toolkit that can handle everything from real-time video previewing to advanced media transformations — and even custom video editing workflows. In this talk, we’ll go beyond playback and explore the hidden capabilities of Media3: - Frame extraction with the new ExperimentalFrameExtractor for high-quality thumbnails and timeline previews - Editing and effects with Transformer and Effect APIs — from speed changes to overlays - Real-time previewing with CompositionPlayer - Understanding when to use preview vs export workflows Whether you’re building a social media app, a sports highlight editor, or simply want to level up your media handling skills, you’ll leave with a toolkit of ideas (and code) to take Media3 far beyond just playback.

Every team eventually hits the same wall: scripts don’t scale, APIs fail unpredictably, and running jobs at scale feels like herding cats. At Smithy, we had to solve this problem while building our opensource security workflow engine, capable of running thousands of workflows per minute across messy, unreliable integrations. In this talk, I’ll walk through how we designed and scaled Smithy’s async orchestration engine using Go and Kubernetes, and what we learned along the way. We’ll cover: * Why async orchestration matters for modern cloud and AI-driven systems. * The challenges we faced: retries, timeouts, state consistency, and observability at scale. * How Go and Kubernetes helped us balance performance with resilience. * AI hooks in practice: where we integrated enrichers and decision logic, and what worked (or didn’t). Demo: a real-world case study of how these design choices power security workflows used today. This is a look under the hood of a system that moved from duct-taped scripts to a production-grade orchestration engine. You’ll walk away with patterns and war stories you can apply to any large-scale cloud or AI workload.

Movies have promised the future where we can talk to AI Assistants with natural language and have them respond with all the nuances of humans. However for years we've been stuck with really "stupid" AI experiences over voice. With the advent of LLMS this is finally a reality! Lots of different technologies have come together to create amazing audio experiences. However, there are some technical problems that we need to solve. In this talk we will learn how to chain these technologies, solve some of the common problems and build a new JARVIS together, powered by the LLM of your choice.

AI has now become part of everyday applications, from recommendations to automated decision-making. But when users don’t understand how or why an AI system makes its choices, trust diminishes and adoption suffers. For developers and designers alike, the challenge is not just building accurate models, but creating experiences where AI feels reliable, explainable, and fair. This talk explores practical UX and design patterns for making AI transparent to users without overwhelming them. I’ll cover strategies such as designing explainable outputs, surfacing confidence levels, handling edge cases gracefully, and giving users control when automation misfires. Attendees will also learn how to pair Google Cloud AI/ML capabilities with front-end design choices that empower users rather than alienate them. By the end, participants will walk away with: - UX principles for designing explainable and trustworthy AI interfaces - Practical techniques for balancing technical complexity with user clarity - Methods of blending AI insights into user journeys to improve trust and adoption - Progressive product education implementation when designing AI systems for users

In this codelab, you'll learn how to build and deploy a Model Context Protocol (MCP) server to extend the capabilities of the Gemini CLI. You will be building godoctor, a Go-based server that provides custom tools for Go development, transforming the Gemini CLI from a general-purpose coding assistant into a specialized Go development expert.

Abstract: Dashboards are great at showing what happened — but what if they could also predict what’s next? In this session, I’ll demonstrate how to integrate AI/ML models directly into BI tools like Tableau, using cloud data warehouses (BigQuery) as the backbone. We’ll cover: Designing an ML model pipeline that feeds into live BI dashboards Avoiding common pitfalls when connecting predictive models to visualisation tools How real-time AI insights change business decisions Attendees will learn step-by-step integration techniques and how to measure the real-world ROI of combining BI and AI. Why it’s unique: Bridges two popular but often siloed skill sets, BI developers and ML engineers into one workflow that’s immediately useful in industry.

As web developers, Chrome DevTools is an indispensable part of our workflow. It’s our second nature to inspect the DOM, set a breakpoint, or check a network request. These actions are so fundamental that they're obvious. But what about your AI coding assistant? For an AI, the browser is a black box. It can help you write code, but it has no way of seeing what that code actually does when it runs. It's programming with a blindfold on. With chrome-devtools-mcp we are bringing the power of Chrome DevTools to AI coding assistants. With it, your agent is able to debug web pages directly in Chrome, dramatically improving the accuracy and effort to identify and fix issues.

This session explores how to develop agents for real-time data, focusing on innovative use cases that aren't your typical chat bot. I will showcase how you can use continuous queries to process and export your data in realtime for consumption by an agent, with human in the loop

Assists Android developers in understanding Android's release timeline and key features, enabling developers to maintain app compatibility.

Discover what's new in Go 1.25 in this practical talk focussed on performance and developer experience. We’ll explore experimental memory arenas for reducing GC overhead, automatic speed boosts from default Profile-Guided Optimisation (PGO), and the new standard version package for cleaner build info.

AI systems are only as good as their guardrails. Without safeguards, they risk hallucinations, biased outputs, or security leaks. This session introduces AI Guardrails and shows how to build them into your cloud-native workflows. We’ll explore validation layers, policy enforcement, and monitoring, using Google Cloud tools and open-source frameworks. Real-world scenarios will highlight how to keep AI systems not just powerful, but also safe and compliant. Audience Takeaway: Attendees will understand the “why and how” of AI guardrails, and walk away with practical techniques to add safety layers to their AI pipelines.

Being 20% more efficient is something you hear about a lot right now. In this talk, I'll go over our strategies for how we achieved these efficiency gains not with AI, but by improving collaboration between designers and developers. This helped us deliver more consistent products and features faster. Our implementation is based on Flutter, but you can apply these learnings to whatever platform you're building with. I will go over: - How building and improving a Design System fits into our product development cycles. - How we ensure the Design System is used efficiently. - How we connect Figma to Flutter. And many more learnings from building and using a Design System.

Ever wondered what it takes to reinvent your career entirely? This isn't your typical 'Bootcamp to job' story. Join Ilias, a former helicopter pilot, as he shares his decade-long journey from navigating the skies to building intelligent applications with Google Cloud and AI. Ten years ago, a career change felt impossible. But driven by a passion for technology, Ilias embarked on a challenging path to learn programming, master cloud technologies, and ultimately become a Google Developer Expert (GDE) in AI and Google Cloud. This session isn't just about what was learned, but how. We'll dive into: The Transferable Skills: Surprisingly, the discipline, problem-solving, and risk assessment honed in aviation proved invaluable in the world of software development. We'll explore these unexpected connections. Navigating the Learning Curve: From initial coding hurdles to deep dives into machine learning, Ilias will share practical strategies for effective re-skilling, focusing on resources and approaches that worked. Building a Community: The importance of finding mentors, contributing to open source, and engaging with the Google Cloud and AI community. A Pilot's View on AI: How a Background in Complex Systems and Critical Decision-making shapes a unique perspective on the potential – and the responsible development – of Artificial Intelligence. Practical Insights into Google Cloud AI: A glimpse into specific Google Cloud AI services and how they can be leveraged for innovative projects. Whether you're a seasoned developer, a tech enthusiast considering a career shift, or simply curious about the power of re-skilling, this talk will inspire you to embrace change and chart your own course in the ever-evolving world of technology. Prepare for a story that proves it's never too late to learn, adapt, and soar!

Large Language Models unlock powerful new capabilities, but they also introduce risks: prompt injection, data leakage, unsafe outputs, and compliance challenges. In this session, we’ll explore Google Cloud’s Model Armor, a new service that proactively screens LLM prompts and responses to ensure safety, privacy, and compliance. I’ll share real-world use cases, architecture patterns, and best practices for integrating Model Armor into enterprise AI systems—whether on GCP, multi-cloud, or hybrid environments. Attendees will leave with practical strategies to safeguard their AI applications and build trustworthy, scalable solutions.

Transform your organisation's data from scattered chaos into AI-powered insights using production-grade MLOps pipelines. This session demonstrates how to orchestrate complex data workflows using GCP Composer, process massive datasets with DataProc, and deploy ML models through Vertex AI pipelines. This talk covers scalable architectures, building end-to-end MLOps pipelines with monitoring and versioning, implementing cost optimisation techniques for large-scale processing, and establishing security best practices for enterprise AI deployments. Attendees will take away pipeline templates and patterns for robust, compliant deployments, aligned with my cloud migration experience

Generative AI is only as powerful as the data that fuels it. In this session, Federico Mestrone explores how data engineering meets AI through modern pipeline design — covering RAG (Retrieval-Augmented Generation), embeddings, and continuous learning patterns. You’ll work on a simple example pipeline at the intersection between ETL and RAG that transforms text data into usable embeddings for a Gemini Enterprise application, while uncovering the growing role of data engineers in shaping the next generation of AI-driven systems.

As we rush to deploy autonomous AI agents in production, we're creating systems with unprecedented impact and decision-making power - from managing support tickets to executing financial transactions and supporting clinical decisions. But, as uncle Ben (Spiderman) wisely told us, with great autonomy comes great responsibility (and potential liability). Recent production failures (Claude Blackmailing, OpenAI suicide case, Grok tweets, Privacy scandals, Replit DB Deletion, Air Canada’s Chatbot fictional policies lawsuit….) underline the principle: if you think safety is expensive, try an accident. Because even if your AI agent won't pass the Turing Test, it might fail your unit tests... and then delete them. Drawing from production deployments, running AI upskilling programmes, and co-authoring the AURA (Agent Autonomy Risk Assessment) framework, this talk presents a practical safety playbook for developers and product teams. We will show how theoretical AI principles can be turned into implementable code patterns and best practices that work with Google's Gemini API and Vertex AI, that every developer needs before shipping their next AI feature. This talk provides immediately actionable patterns that prevent your AI agent from becoming the next cautionary tale, preventing Murphy’s Law to turn into full Age of Ultron style.

Have a great app idea but don't know where to start? This session will guide you through a powerful, hybrid workflow to transform your concept into a real application. We'll explore a unique three-part process that combines the creative power of Gemini, the organisational clarity of NotebookLM, and the structured development environment of Firebase Studio. First, you'll learn how to leverage Gemini as your creative co-pilot to brainstorm features, refine your concept, and explore potential user flows. Next, we'll show you how to use NotebookLM to document, organise, and synthesise these ideas into a comprehensive project plan, helping you structure your thoughts and keep everything in one place. Finally, you’ll see how Firebase Studio provides a streamlined pathway to bring your app to life, allowing you to build out the core functionality while maintaining crucial control over key design decisions, including the user experience (UX) and the underlying data models. This session can be delivered as a 30 minute talk, a 45 minute talk/demo or a longer interactive talk/demo/workshop.