SmartPicks

Inspiration

PrizePicks inspired us because it gives users access to live player data for decision-making. We wanted to take that idea further by allowing users to ask natural-language questions and get real-time, AI-generated insights about player performance.

What it does

SmartPicks is an AI sports analyst that answers user questions about NBA players using real-time statistics. It retrieves player data from public APIs and uses Google’s Gemini model through Vertex AI to generate concise, data-driven insights and recommendations.

How we built it

We built the interface in Streamlit, integrated the nba_api library to fetch live player stats, and used the Vertex AI Python SDK to connect to Gemini 1.5 Flash. Prompts were designed to combine user queries with recent player data for context-aware responses.

Challenges we ran into

Accessing consistent real-time data was challenging due to API limitations. Extracting player names accurately from natural text required additional logic, and tuning prompts for concise, relevant responses took iteration.

Accomplishments that we’re proud of

We successfully combined live NBA statistics with generative AI analysis in real time, created a working conversational interface, and built a clean, extensible codebase that could easily expand to other sports.

What we learned

We learned how to integrate structured data with generative reasoning, use Vertex AI effectively, and design prompts that produce analytical yet concise outputs. We also gained experience optimizing API workflows and latency in Streamlit apps.

What's next for SmartPicks

Next, we plan to add multi-player comparisons, visual summaries using Gemini Image, expand to other sports, and host the app publicly so users can get live insights during games.

Built With

• cloud
• google