​Build with AI: San Francisco Edition Day 1

​Please also sign up for the second day: https://luma.com/op7obu11

​Agenda: Saturday, March 21

​09:00 AM | Kickoff Check-in & Registration

​10:00 AM | Hackathon Challenges, Present Judging criteria

​11:10 AM | RocketRide Demo integrated with Antigravity and Challenge presentation by Aaron Jimenez, RocketRide

​10:30 AM | Building a Multimodal Multi-Agent Discovery System by Annie Wang, Google. In this workshop, you will build an orchestrated AI application featuring multi-turn image generation and a multi-agent processing system. You will begin by using the Gemini Image API via the Python SDK to generate and refine a consistent user avatar through multi-turn prompt engineering. Next, you will develop a parallel multi-agent system to analyze complex multimodal data (images, video, and audio). You will learn to implement two Model Context Protocol (MCP) patterns: building Custom MCP servers deployed on Google Cloud Run for domain-specific AI logic, and connecting to Google's managed OneMCP for BigQuery integration. By orchestrating these agents, your system will analyze environmental evidence to deduce a location and register the user on a live global map.

​​11:40 AM | Building Production-Ready AI Agents: From Inference to Deployment by Roan Weigert, AI Engineer & Featured Speaker. In this workshop, you will explore the core engineering principles behind deploying AI agents in production, including agentic workflow orchestration, KV cache mechanics, model quantization, and real-time latency optimization for multi-agent systems.

​12:00 PM | Use Gemini in Vertex AI as an AI Director to watch and edit videos by Anna Nerezova, GDG Lead NYC

​12:30 PM | Lunch Connect with fellow builders

​01:30 PM |Building an ADK Bi-Directional Live Agent by Christina Lin. In this hands-on lab, you will build a full-stack, real-time multimodal AI application with an Agent. You will develop a React frontend to capture live video and audio, and a Python backend using Google's ADK to process this data. By establishing a bi-directional WebSocket connection, you will stream the media to the Gemini Live API, which will interpret hand gestures to execute specific commands.

​3:00 PM | Group finding and pitching of ideas. Onboarding of Google Cloud credits for hackathon, Onboarding for special challenge for RocketRide. Official start of Hackathon with Groups mentored by GDES and trainers.

​03:30 PM | Hacking

​06:00 PM | Day 1 Wrap-Up Final logistical announcements as we transition into the second day.

​🏆Register to win prizes!🏆

​
Prizes include:
💰 $2,500+ Cash prizes
🎤 Opportunity to pitch to the AI Futures Fund for the winning team
🎟️ Google I/O tickets for the winning team

Notion Access: https://sly-salary-8bc.notion.site/Build-with-AI-SF-Hackathon-31ba4bb59d7180a9a03cddb43787b78e