8:00 AM | DOORS OPEN | |
9:00 AM | INTRO TO LARGE LANGUAGE MODELS WITH BARD AND PALM by Luiz GUStavo Martins | Large Language Models (LLMs) are revolutionizing the way we interact with computers. On this talk, let's learn about Google new suite of that make it easier for developers to build on top of our best LLMs. These tools and APIs will allow developers to create applications that can comprehend and generate natural language with a level of nuance and fluidity that was never before possible. |
9:30 AM | UNLEASHING AI MUSIC COMPOSITION: EXPLORING MUSICLM by Nina Ma | In this 25-minute talk, I will delve into the realm of AI music composition tools, with a specific focus on MusicLM by Google. MusicLM is a cutting-edge model that generates high-fidelity music from text descriptions, enabling musicians and creators to effortlessly translate their creative visions into captivating melodies. Through this talk, attendees will gain insights into the capabilities of MusicLM and its potential impact on the music industry and creative processes.
1. Introduction to AI Music Composition
- Overview of conditional neural audio generation and its applications
- Challenges in generating rich audio sequences from text descriptions
2. MusicLM: Unleashing AI Creativity
- Hierarchical sequence-to-sequence modeling approach
- High-fidelity music generation at 24 kHz with consistency
- Comparison with previous systems in audio quality and adherence to text descriptions
3. Expanding Boundaries: Beyond Text Conditioning
- Incorporating additional melody inputs for personalized music generation
- Flexibility and versatility of MusicLM's conditioning capabilities
4. Introducing MusicCaps: A Groundbreaking Dataset
- Curated collection of 5.5k music-text pairs prepared by expert musicians
- Supporting evaluation and future research in AI music composition
5. Evaluation, Findings, and Responsible Development
- Metrics and comparisons with previous systems
- Addressing concerns of memorization and biases
6. Future Directions and Broader Impact
- Exploring lyrics generation, improved conditioning, and vocal quality
- Mitigating risks and responsible use of AI music composition tools |
10:00 AM | MAKING YOUR FLUTTER APPS SHINE WITH SHADERS by Raouf Rahiche | Flutter has supported fragment shaders for approximately a year, yet their usage remains relatively low. This lack of adoption largely stems from the difficulty of getting started and the initial challenge of identifying relevant use-cases. In this talk, we will delve into the world of shaders. By exploring their various applications and implementing some engaging, practical features, we aim to illuminate their potential and make them more accessible to developers. |
10:30 AM | GOOGLE CLOUD GENERATIVE AI STORY by Sadeeq Akintola | High level overview of the GCP Generative AI narrative |
11:00 AM | USING THE PLATFORM - WHY YOU DON'T NEED ANY WEB FRAMEWORKS IN 2023 by Akash Joshi | After my experience working with multiple companies over the past few years, and building several open-source projects - I've realised that most solutions for building websites are over-engineered, and the web platform already provides us with a lot of fundamentals for deploying apps to the web, and also mobile.
Table of Contents:
1. What is the platform?
2. Basics - Replacing jQuery
3. Getting device data - location, battery, accelerometer, etc on the web
4. Web Components - no need for JS frameworks
5. PWAs and TWAs - how to deploy web apps to the app store |
11:30 AM | EXPLORING ML MODEL APPLICATION ON LIVE IMAGE STREAMS IN FLUTTER: PAST, PRESENT, AND FUTURE by Vadym Pinchuk | This topic focuses on the practical implementation of machine learning (ML) models on live image streams using the Flutter framework. We will delve into the existing solution with the TensorFlowLite (TFLite) plugin, which allows ML model application for object detection on image. By examining the current state of the technology, we will assess the pros and cons of this approach and discuss potential applications.
Furthermore, we will explore the future prospects enabled by the upcoming MediaPipe plugin, an anticipated framework by Google. MediaPipe facilitates on-device ML by providing powerful real-time processing and analysis of live video streams in Flutter applications. It handles the underlying logic with native implementations for Android and iOS platforms, what will make it very performant, comparing to current state.
Throughout the session, attendees will gain insights into the opportunities presented by ML on live image streams in Flutter development. By understanding the past, current state, and future possibilities, developers will be equipped to harness the full potential of ML model integration in Flutter for object detection and image stream processing applications. |
12:00 PM | LUNCH | |
1:00 PM | JETPACK COMPOSE FOR MAPS by Darryl Bayliss | Jetpack Compose is taking the Android development world by storm. With plenty of apps dedicated to showing information via maps, how does Jetpack Compose provide support?
This session aims to give attendees interested in mapping solutions the answer to that question in ~25 minutes, by exploring the Google Maps Compose Library, how it works in conjunction with acquiring permissions to make the most of the map and what benefits it provides over the old View based mapping approach.
The structure of the session would be as follows:
1. A brief recap of the mapping space, what apps are in the space and some metrics to show the reliance people have on mapping apps
2. A brief recap of Jetpack Compose, and the benefits it brings via it's Declarative UI approach.
3. A look into how the Maps Compose Library from Google works, and what advantages it brings.
4. Showing best practices using the Maps Compose Library, in collaboration with acquring the correct permissions for the map to function
5. A summary of the topic, to recap key points and give attendees a chance to ask questions. |
1:30 PM | WASM AND THE NEW PERKS OF FLUTTER WEB by Djamaleddine Belilet | WASM is a new technology that allows developers to compile code from other languages, such as C++ and Rust, to a format that can be run in the browser to deliver high-performance web applications run in a secure sandbox.
WebAssembly (WASM) is emerging into the Flutter ecosystem and it will become increasingly popular in the future.
In this session, we will discuss the benefits of using WASM to build high-performance web applications as well as some of the challenges of using WASM, such as the lack of mature tooling and the small community of developers who are currently using the technology.
By the end of this talk, we will have gained a clear understanding of how to incorporate WebAssembly (WASM) into Flutter web apps, its benefits and pitfalls, as well as some of the new features that have been released with Flutter (Web) 3.10. |
2:00 PM | UNLEASH THE POTENTIAL OF GOOGLE WORKSPACE & GOOGLE DRIVE by Jasper Duizendstra | Delve into the possibilities of enhancing and automating Google Drive's existing functionalities using Apps Script. During this session, we'll discuss the conception of a document publishing add-on, the generation of custom reports, and the automation of key tasks, all aimed at refining your document management process. We'll also touch upon the role of Google Chat in facilitating efficient document searching.
Key Takeaways:
- Strategies for designing an efficient document management system using Apps Script.
- Leveraging and automating Google Drive's inherent functionalities for optimal document management.
- Considerations for system implementation using various tools, with a spotlight on Apps Script. |
2:30 PM | EXPLORING THE ART OF MLOPS FOR DEPLOYING LARGE LANGUAGE MODELS USING VERTEX AI by Nishi Ajmera, Srivalsan Mannoor Sudhagar | This talk will discuss in detail about the need of MLOps in the current ecosystem and how MLOps can increase the release to market . This will also discuss in detail about how to leverage capabilities in Google Vertex AI to implement State of the Art MLOps pipelines to deploy Large Language Models in Google Cloud. |