Back
AI Meeting Assistant
Year
2024
Tech & Technique
OpenAI, RAG, FAISS, Whisper, Python
Description
A desktop assistant that listens to live meetings, detects questions as they are asked, and answers them from your own documents.
Key Features
- Live Audio Capture Captures and transcribes system audio during meetings
- Question Detection Spots questions in the live transcript automatically
- RAG Grounding Answers from a FAISS index built out of your uploaded files
- Context You Can Change Upload or delete files at runtime to reshape the answer context on the fly
My Role
AI Engineer
Sole developer:
- Built the chunking and FAISS retrieval layer over uploaded documents
- Built the live system audio capture and Whisper transcription pipeline
- Built question detection and the answering flow over OpenAI, grounded in retrieved context
- Implemented uploading and deleting several files at runtime to reshape context during a session