• Collaborated on a real-time pipeline converting speech prompts into 3D models, images, videos, or GIFs for holographic display.
Designed a modular middleware that performed speech-to-text, API-based intent detection, and routed requests to appropriate
generative AI tools.
• Integrated Unity via WebSocket to dynamically load and display assets at runtime in holographic scene based on file type.
• Developed the intent detection pipeline to classify user prompts and determine the desired content type and associated features,
generating structured JSON outputs via LLMs (ChatGPT, Gemini, etc) using YAML templates to enable automated asset
generation through API calls.