| Breakfast and Registration | Lite bites and networking |
| Welcome by Community Members | Brief intro to the agenda and warm welcome from volunteers. |
| Keynote by Chris Perry: Enabling the next billion developers | Discussion of changing landscape of developer tooling and how we can enable more people to build. |
| Keynote by Jeremy Nixon: AI Powered Scientific Research: AI for Biology, Physics, Mathematics, and AI Itself. | From AlphaFold to FunSearch, the automation of research progress with artificial intelligence is a clear frontier for the core problems humanity faces. Nobel prizes do not unveil the deeper lesson - parallelizable automated knowledge discovery will transform scientific research deeply, perhaps beyond recognition. Let’s examine the consequences of the latest paradigmatic AI progress for scientific research. |
| Track 1 (Talk): Chrome's Built-In AI by Hemanth HM -- Track 2(Workshop): Intro to workshop / Building blocks for hackathons by Sako M | Track 1 (Talk) Description:
This session delves into the Prompt API, an early-stage proposal by the Chrome built-in AI team, designed to address specific developer challenges. Hemanth will provide live demonstrations of the API, highlighting its capabilities and potential applications. Additionally, he'll explore upcoming API proposals related to window.ai, discussing their technical details and implications. This session is ideal for developers and tech enthusiasts interested in the latest advancements in web technologies. Attendees will have the opportunity to provide valuable feedback and contribute to the future development of these APIs. h3manth.com/ai/chromeai
-----------------------------
Track 2 (Workshop) Description:
You’ll learn how to set up a high-performance local environment, use 'aiproxy' for efficient proxy traffic management, and automate dependency updates with PlatformGen. Topics to be covered include: Setting up local environment for speed, Leveraging aiproxy to handle proxy traffic, and Apply PlatformGen - Autonomous Dependency Manager
|
| Track 1 (Talk): Unlocking the full potential of Vertex AI Gemini: A developer's guide to prompt engineering by Rajesh Thallman and Abhishek Bhagwat -- Track 2 (Workshop): Human-in-the-loop agents with assistant-ui by Simon Farshid | Track 1 (Talk) Description:
Want to build powerful and reliable applications with Vertex AI Gemini API? Deep dive into the art of effective prompting with Gemini models. Join our experts to gain practical strategies for crafting effective prompts, maximizing model performance, and overcoming common challenges.
-----------------------------
Track 2 (Workshop) Description:
Learn about the different patterns to incorporate human feedback in your AI agents. We will go through a hands on live coding session to build an interactive AI agent using assistant-ui, an open-source React component library for chat interfaces.
Topics covered will include:
* Understanding Human-in-the-Loop: An overview of why human feedback is essential in AI development and how it can be systematically incorporated, including Generative UI, approvals and more.
* Design Patterns: Learn about the right UX patterns for human input
* Building with assistant-ui: Step-by-step guidance on using assistant-ui to develop interactive agents.
|
| Lunch Break | Nom, nom and networking |
| Track 1 (Talk): Breaking into Singapore as a Startup by Ed Choudhry -- Track 2 (Workshop): Solving the Python-Javascript dilemma with Reflex--a modern OSS Python web framework by Elvis Kahoro | Track 1 (Talk) Description:
Recently landing back in town from a successful pilot that has opened the door to multiple customers, Ed Choudhry, CEO and CoFounder of Barricade AI would like to share a special opportunity for top talent and founders to explore working with the Singaporean Government and invite submissions to the recently posted Challenge Statements. Ed will provide his personal experience since the summer and provide insights how you can be the next to build your next successful venture in the APAC region.
-----------------------------
Track 2 (Workshop) Description:
Reflex is an open-source (Apache 2.0 & 20k+ GitHub stars) framework empowering Python developers to build internal (data, AI, and web) apps faster - with no JavaScript required. Build both your frontend and backend in a single language, Python. (pip install reflex). In this workshop, we'll go over how to set up Reflex and start building a dashboard app. |
| Track 1 (Talk): Transforming Enterprise Productivity with LLMs and Document AI by Lucy Park -- Track 2 (Workshop): Build e-commerce recommendations, search and analytics with vectors by Daniel Svonava | Track 1 (Talk) Description:
In the enterprise landscape, true productivity gains with AI go beyond just acquiring tools—it’s about uniting them to tackle complex challenges. In this session, we introduce: (1) Solar LLM, a powerful, purpose-built language model tailored to meet the unique demands of modern enterprises; (2) Document Parse, a solution designed to seamlessly transform complex documents into AI-readable formats; and (3) inspiring, real-world examples showcasing enterprise-level integration. Discover how these technologies work in harmony to automate, accelerate, and elevate workflows, unlocking new levels of efficiency and insight across organizations.
-----------------------------
Track 2 (Workshop): Build e-commerce recommendations, search and analytics with vectors by Daniel Svonava
Description:
use your unstructured AND structured data to handle queries like "popular blue t-shirts around $50", for building a time personalized shopping feed and other features useful for e-commerce businesses - with a few lines of Python, running at scale in your own GCP account.
|
| Track 1 (Talk): Powerful Google Developer Tools for Immediate Impact by Wesley Chun -- Track 2 (Workshop): Get Chunked! by Adam Chan | Track 1 (Talk) Description: Are you a professional developer or student new to Google developer tools? You may know Google for search, YouTube, Android, Chrome, and Gmail, but did you know Google has many programmer resources like compute & storage services, networking & security, APIs, AI/ML tools, and hosting platforms? This session introduces you to some of the most impactful services with the purpose of inspiring what you can accomplish in a short amount of time. They are great for work projects on your plate or for student coursework or hackathons.
Highlights include where to run your apps in the cloud without managing VMs (serverless), different types of databases to store your data, "building block" AI/ML APIs to help get you started even if you have little or no experience with machine learning, and general-purpose Google APIs giving you access to resources like spreadsheets, big data analysis tools, and so on. The session concludes with a variety of applications demonstrating the versatility and usefulness of Google APIs to show you what is possible. Join us to discover how Google can help you write your next awesome app!
-----------------------------
Track 2 (Workshop) Description: Join this multistep challenge presented by Adam at Weaviate that will take you from zero to hero on chunking strategies for retrieval powered applications that leverage large language models. |
| Track 1 (Talk): Building a gen AI MVP in 8 weeks by Mariane Bekker -- Track 2 (Workshop): Low Latency on LLMs and Agent Workflows by Jose Menendez | Track 1 (Talk) Description:
We will go through simple steps to get your AI MVP up and running in 8 weeks or less
-----------------------------
Track 2 (Workshop) Description: Low latency on LLMs allow for some fun agentic workflows! |
| Track 1 (Talk): Communicating Technical Products to a Nontechnical Audience by Erika Bahr -- Track 2(Workshop): Harnessing LLM to Power Chatbot developing with Code-generation and Your GenAI App Production-Ready with Tracing & Evaluations by Sam Stowers | Track 1 (Talk) Description: Has an executive ever cut your highly valuable project? Effectively conveying the value of technical products to nontechnical stakeholders is crucial to fostering understanding, alignment, and adoption across organizations. In this session, Erika Bahr, CEO of Daxe, will share strategies for explaining complex AI solutions to audiences ranging from product managers and directors to clients. Drawing from her experience with Fortune 500 companies and her work in AI go-to-market strategy at AWS, Erika will provide practical insights on translating intricate technical details into clear, impactful messages. Attendees will learn approaches for achieving buy-in and connecting with diverse stakeholders on advanced technologies that drive business transformation.
-----------------------------
Track 2 (Workshop) Description:
LLM demos are easy - making them reliable enough for production is hard. Weave (from Weights & Biases) helps with every stage of development and deployment, so you can iterate rapidly and build confidence in your application.
Understand complex queries with tracing (3 lines of code!!)
Rapidly identify cost & latency bottlenecks
Implement evaluations to make LLM responses consistently good
Test which LLMs are best for your needs.
|
| Track 1 (Talk): LLMinABox: On Device Personalized Diary And Concierge using your voice and Gemma by Rabimba Karanjai -- Track 2 (Workshop): Generating Interactive Environments for Synthetic Training Data with Google Cloud and Diffusion World Models by Luke Hollis | Track 1 (Talk) Description: The talk focuses on a cutting-edge AI-powered solution that leverages large language models (LLMs) and on-device processing to enhance productivity and learning.
Here's a summary of the key features and benefits:
- Effortless Voice Capture: Transcribe and organize thoughts, ideas, and meeting notes using natural language voice interaction.
- AI-Powered Meeting Summaries & Action Items: Automatically generate concise summaries and extract actionable items from meetings.
- Personalized Learning Tools: Create tailored study materials, code examples, and explanations based on voice recordings, meetings, or uploaded notes.
- On-Device Processing: Ensure complete privacy and security by keeping all data on the user's device.
---------------------
Track 2 (Workshop) Description:
Inspired by "Genie: Generative Interactive Environments" (Bruce et al., 2024), this hands-on workshop
introduces you to generating interactive environments to build simulations or synthetic training data using diffusion world models. We'll demonstrate the basics of how world models can create and control interactive scenes, starting with simple Atari games. You'll learn the fundamental concepts behind
generating synthetic training data while getting practical experience with video encoders.
Building on real examples from robotic manipulation tasks, you'll learn how this method can provide diverse training data without requiring expensive real-world data collection, even generating physically plausible interactions with deformable objects. Based on Alonso et al., 2024, Bruce et al., 2024, Hafner et al., 2024, and other recent research. You'll leave with a working Colab that you can build on for your own projects. |
| Track 1 (Talk): by Arjun Chintapalli -- Track 2 (Workshop) Harnessing LLM to Power Chatbot Development with Code-generation and Doc-explanation by Huan Li | Track 1 (Talk): Demo of rtrvr.ai chrome extension
Description:
Demo of rtrvr.ai chrome extension that can do tasks autonomously on the web and extract structured data. As well as dive deep in the technical architecture of using Google Cloud, Firebase, Gemini, and Chrome API's
-----------------------------
Track 2 (Workshop) Description: Conversational AI has empowered the Chatbot application in recent years, making the Conversational User Interface (CUI) increasingly valuable. WhatsApp & WeChat are two of the largest world IM platforms.
What if we could turn our personal accounts into a chatbot? It will be precious for many use cases if we can turn our account into a chatbot as a practical and convenient CUI application to serve users.
In this workshop, you will learn how to use LLM to assistant you to convert your requirement in natural language to source code using Wechaty SDK.
1/Learn what Wechaty SDK can do
2/design your requirement for automating your WhatsApp personal account
3/generate source code by prompting your requirement in natural language to LLM |
| Wrap up and Networking | Decompress and Connect. |
