I had the amazing opportunity to attend the Google Build with AI 2025 event, hosted by GDG Sri Lanka at WSO2, and it was nothing short of inspiring. As someone deeply passionate about tech and AI, being invited to both days of this event was an absolute honor! 🤩
From deep technical sessions to real-world applications of AI, fun interactions, hands-on moments, and spontaneous team activities — this event had it all. Here's a breakdown of my experience over the two eventful days.
📅 Day 1: Diving Deep Into AI Fundamentals and Responsible Systems (May 17, 2025)
🔧 Creating Cutting-Edge AI Apps with Firebase Studio and Genkit
Speaker: Vrijraj Singh
Vrijraj kicked off the event with an energetic session on how to build modern AI applications using Firebase Studio and Genkit. He showcased how easy it is to connect GenAI models with Firebase services like Firestore, Functions, and Authentication — accelerating development while keeping things scalable and secure.
🤖 Building Connected Agents: An Overview of Google’s ADK and A2A Protocol
Speaker: Suresh Peiris
Suresh walked us through Google’s Agent Development Kit (ADK) and the Agent-to-Agent (A2A) protocol. His talk emphasized the importance of connected agents that can communicate, coordinate, and solve complex tasks together. A real eye-opener for those building collaborative AI systems.
⚡ Optimising Gemma for Inference
Speaker: Thiru Dinesh
Thiru explained how to optimize Google’s lightweight open-source model Gemma for fast inference on various platforms. He shared valuable insights about quantization, fine-tuning, and deployment strategies — all crucial for efficient edge AI applications.
🧠 Redefine LLMOps - with a Robust Dual Memory Architecture
Speaker: Isuru Alagiyawanna
This was one of the most technically rich sessions. Isuru introduced a novel Dual Memory Architecture for LLMOps, blending short-term and long-term memory concepts to manage AI workloads more efficiently and make model interactions more contextually aware. Definitely a future-ready idea!
🔐 Secure and Governed AI: Building Responsible Models
Speaker: Roy Ian
Roy’s session brought the focus to AI ethics, security, and governance. He talked about best practices for building AI that is secure, explainable, and fair. It was a powerful reminder that innovation must always be matched with responsibility.
🎤 Panel Discussion
The day ended with a panel discussion featuring all the speakers. The panel covered future trends in AI, challenges in responsible adoption, and thoughts on how GenAI will evolve in the next 3–5 years. It was engaging, insightful, and thought-provoking.
📅 Day 2: Advanced Agentic Thinking and Practical Applications (May 18, 2025)
☁️ Let’s Talk to Your Cluster: LLM Agents Powered by Kubernetes MCP Server
Speaker: Chamod Perera
Chamod showed how to build LLM agents that directly communicate with Kubernetes clusters through MCP Server. This session merged DevOps and AI beautifully, enabling use cases like autonomous server diagnostics and recovery actions through natural language.
🛠️ Beyond Runbooks: AI Agents as Your Autonomous Day 2 Operators
Speaker: Isala Piyarisi
Isala tackled a real business problem — Day 2 Operations — and how they can be managed by autonomous agents. From handling alerts to performing remediation tasks, the agent framework he presented can help reduce manual workload in ops teams drastically.
🧩 Agentic Reasoning with GraphRAG
Speaker: Nirodya Pussadeniya
Nirodya introduced us to GraphRAG, a technique that enables agents to reason more logically by traversing knowledge graphs. The combination of Retrieval-Augmented Generation and graph databases was brilliantly explained and had a lot of real-world relevance.
🧬 Looking Inside AI Brains: How to Control What Neural Networks Think
Speaker: Keshan Sodimana
Keshan closed the day with a futuristic talk on AI interpretability. He discussed how neural networks form thoughts, and how we can influence or guide their inner workings. From attention heads to activation mapping, this was truly mind-blowing!
🏆 Winning the Bottle – A Fun Bonus! ✨
In the middle of all the intense learning, there was a surprise mini-competition! 🎯 They asked attendees to form random teams and brainstorm an AI-based business solution to solve a real-world problem.
Our team of 3 (randomly matched!) came up with an idea inspired by one of our teammate’s research topics — a solution involving Dialog signal towers 📶. The concept was to use AI to predict and optimize tower load distribution to reduce signal issues and improve user experience, especially in rural or high-traffic areas.
We submit idea… and boom! it was best idea among other submits and We selected as Winners. Our Team members received some swags; I won a bottle! 🏆 Just a bottle? Yes. But it felt like an Oscar 😂.
🎉 Overall Vibe: Interactive, Fun, and Full of Energy!
Besides the sessions, the entire event was filled with fun interactions, networking moments, and casual Q&As with the speakers. Everyone—from organizers to attendees—was warm, welcoming, and buzzing with ideas. It was a beautiful mix of tech and community.
Thanks again to GDG Sri Lanka and WSO2 for hosting such an incredible event. If this is where AI is headed, count me all in! 💪🤖✨