Hi, I'm Matthew Peng.

Software engineer focused on infrastructure, distributed systems, and AI applications. I build production systems spanning data infrastructure, developer tooling, and LLM/RAG applications.

Projects

Lighthouse Church Chatbot

ChatLCC

Production RAG & Sermon Analytics Platform

Try ChatLCC

Built and deployed a full-stack RAG system providing semantic search and structured analytics across 2,028 sermons and 46,000+ vector records. Uses LangChain to route queries between vector retrieval and MySQL-backed metadata tools, enabling semantic questions, sermon counts, pastor filtering, and date-range analysis.

Hours of sermons
1,500
Sermon recordings
2,028
Searchable vector records
46K+

Architecture

Requests flow from the user through Next.js and FastAPI to LangChain, which orchestrates retrieval, metadata tools, and generation.

Requests flow from User to Next.js to FastAPI to LangChain. LangChain connects to Pinecone for semantic retrieval, MySQL for structured metadata, and an LLM for generation and citations.

Engineering highlights

  • Query rewriting & retrieval

    Rewrite questions for semantic retrieval over the sermon archive using Pinecone.

  • Structured sermon analytics

    Use MySQL-backed metadata tools to count sermons and filter by pastor or date range.

  • Grounded responses

    Combine validation, source citations, and fallbacks to support reliable answers.