AI Meeting Transcription, Summarization & Q&A System (RAG + LLM):

Objective

Built an end-to-end system for meeting transcription, summarization, and context-aware Q&A using Whisper, Qdrant, and LLMs, with FastAPI + React for real-time processing and interactive querying.

Stack

Delivery highlights

• Developed an end-to-end meeting intelligence system that processes uploaded audio/video by splitting recordings, transcribing speech using Whisper, and generating structured Thai summaries and action items with GPT-4o-mini, while storing transcript embeddings using OpenAI text-embedding-3-small in Qdrant for semantic retrieval; the system enables context-aware Q&A by retrieving relevant transcript segments, returns source-backed answers with traceable references, tracks processing status in real time, and provides a React-based interface for reviewing transcripts, summaries, audio segments, and downloadable reports.