Science Bro - The Netherland's Shortest Hackathon
AI Tinkerers - Amsterdam
Hackathon Showcase

Science Bro

Team led by a PhD student at Maastricht University specializing in audio ML; Python/PyTorch developer of ACES, torchvggish and open_clip—open-source audio/vision expertise.

1 member

PodSearch: AI-Powered Podcast Discovery & Insight Engine

PodSearch is a groundbreaking MCP (Machine Callable Program) ecosystem that transforms how we interact with podcast content. This innovative platform seamlessly integrates podcast discovery, audio processing, and advanced transcription capabilities into a unified, AI-ready interface.

Core Capabilities:

  • Intelligent Podcast Discovery: Search the vast iTunes podcast library with precision to find exactly the content you’re looking for
  • Seamless Audio Processing: Download, cache, and optimize podcast audio with automatic format conversion and smart bandwidth management
  • State-of-the-Art Transcription: Transform spoken content into searchable text using Groq’s Whisper LLM for unparalleled accuracy
  • Semantic Search: Find specific moments across your podcast library with advanced vector-based semantic search capabilities

Technical Excellence:

Built with a modular architecture featuring three powerful MCPs:

  • podcast-search: Podcast discovery and RSS feed processing
  • podcast-audio: Efficient audio downloading, caching, and optimization
  • podcast-transcript: Advanced transcription with ChromaDB vector storage for semantic search

The system elegantly handles everything from initial podcast discovery to final searchable transcripts, making podcast content as accessible and useful as written text. The architecture showcases modern best practices with asynchronous processing, efficient resource management, and robust error handling.

PodSearch represents the future of audio content interaction - where the wealth of knowledge in podcasts becomes as discoverable, searchable and valuable as any text-based resource.

AI Tinkerers Groq