Inspiration# About the Project: MLB Superfans
Inspiration
Watching MLB games, I noticed that fan engagement was often limited to static statistics and delayed updates. I was inspired to create a platform that brings real-time insights, interactive analytics, and dynamic conversational experiences directly to fans. The opportunity to leverage Google Cloud’s advanced AI services—such as Vertex AI for predictive analytics and Gemini models for conversational AI—sparked the idea for MLB Superfans. I envisioned a system where fans could not only follow live game data but also receive personalized commentary and predictions, creating a more immersive and engaging experience.
What I Learned
Building MLB Superfans was a transformative learning experience. I deepened my understanding of cloud-based data processing and real-time analytics using Google Cloud Pub/Sub and Dataflow. Working with Vertex AI taught me how to train and deploy machine learning models effectively, while integrating Gemini models helped me explore the nuances of natural language processing and conversational AI. The project also honed my skills in mobile development with React Native, emphasizing the importance of scalability, user experience, and seamless API integration.
How I Built the Project
- Data Pipeline: I set up a live data ingestion system using Cloud Pub/Sub to stream MLB game data, processed via Dataflow.
- Predictive Modeling: Historical game data was used to train predictive models with Vertex AI, which were then deployed as REST endpoints.
- Conversational AI: I fine-tuned and deployed Gemini models to generate engaging, context-aware commentary and integrate them into a chatbot interface.
- Mobile App Development: The React Native app was developed to serve as the user interface, displaying live statistics, predictions, and chatbot interactions in real time.
- Integration: All components were integrated into a cohesive platform that delivers a seamless, interactive fan experience.
Challenges Faced
- Data Acquisition: Accessing reliable live MLB data was challenging, so I developed simulated data streams for initial testing.
- Model Tuning: Achieving high accuracy with predictive models required multiple iterations and careful fine-tuning on Vertex AI.
- System Integration: Coordinating real-time data flow between cloud services, AI models, and the mobile app demanded robust API design and thorough testing.
- Performance Optimization: Minimizing latency in processing and updating live data was a significant challenge, addressed through optimization of data pipelines and API calls.
Log in or sign up for Devpost to join the conversation.