A simple RAG-powered chatbot and knowledge base trained on curated HR, CV, and career advice threads.
๐ Project Overview
Role: Full-Stack Developer
Stack: Next.js, React, TypeScript, Tailwind CSS, Node.js, Express, Qdrant, OpenAI API
Timeline: 1 week
๐ What is HALO?
HALO is a conversational assistant built around a real career specialist's advice. Instead of giving generic ChatGPT answers, it references a custom dataset of tweets, threads, and hiring guides to answer questions about CV optimization, job search strategies, and interview prep.

HALO
โ๏ธ How It Works
Data Ingestion: A script cleans up raw Twitter threads (removes noise, short replies, and links) and stitches multi-part tweets into full, readable posts.
Vector Search (RAG): Posts are converted to embeddings with OpenAI (
text-embedding-3-small) and stored in a Qdrant vector database.Grounded Answers: When a user asks a question, HALO retrieves the most relevant posts and feeds them as context into GPT-4o, streaming the response back in real time with source citations.

HALO - Live Intelligence
๐ฅ๏ธ What I Built
Chat Interface: Clean, distraction-free chat with real-time text streaming, suggested questions, and clickable citation cards showing the exact source posts.
Insights Feed: A simple browsable directory where users can read curated career tips and frameworks without having to chat.
Responsive UI: Fast-loading interface with full light/dark mode support built in Tailwind CSS.

HALO - Chat Interface
๐ก Key Takeaway
A straightforward project showing how to turn fragmented social content or notes into a useful, grounded AI assistant using standard RAG patterns and a clean UI.
0 likes ยท 1 view ยท Nov 20, 2026
Built HALO, a full-stack RAG assistant and career knowledge base. Built the responsive Next.js frontend, Express API and Twitter thread ingestion pipeline.