AI Chatbot (RAG) for Porfolio

AI Chatbot (RAG) for Portfolio

24/7 site assistant that answers from my resume, LinkedIn & portfolio — grounded in real sources with citations and guardrails.

AI/LLM RAG n8n Supabase OpenAI

Role

Product strategy & UX
RAG architecture, chunking & prompting
Implementation (n8n + OpenAI + Supabase, Claude Haiku for coding assist)

Highlights

Ingests resume, LinkedIn & portfolio, auto-splitting into 500-character “chunks” of text.
OpenAI embeddings + Supabase pgvector for semantic “meaning” search instead of keywords.
Two-workflow design: real-time Q&A plus ingestion & hourly embedding backfill.
Guardrails for hostile/off-topic queries and a drop-in JS widget with three-dot “thinking” indicator.

Try the Chatbot View Code Watch Walkthrough

Chatbot answering with source chips — Grounded answers with sources from my resume, LinkedIn & portfolio.

RAG architecture: query → embeddings → Supabase → OpenAI → response — RAG flow: embed query → retrieve similar chunks → grounded completion.