Ask Haseeb AI is a retrieval-augmented generation (RAG) assistant built by Haseeb Sagheer. It ingests documents from Google Drive, stores embeddings in Pinecone, and answers questions through a FastAPI backend and a chat interface. It is live at ask.haseebsagheer.com and the code is public.
Live means you can use it now at ask.haseebsagheer.com. It was first built in 2025 and rebuilt in October 2026 with a new interface, streaming answers and a new knowledge base. The source code is public on GitHub, and there is a recorded walkthrough of the first version further down this page.
Ask Haseeb AI is an assistant that answers questions about my skills, projects and experience. It does not rely on what a language model happens to know. It retrieves the relevant parts of my own documents and writes the answer from those.
Documents live in a Google Drive folder. When a new file appears, the system picks it up, cleans it, splits it into overlapping chunks of about 1,000 characters, embeds each chunk, and stores the vectors in Pinecone. A question is embedded the same way, the closest chunks are retrieved, and an OpenAI chat model writes the answer from that context.
Around that sits a FastAPI backend with REST endpoints, a simple chat front end, and a production deployment on a VPS behind Nginx with HTTPS. It is the project where I built every layer of a retrieval system myself.
Recruiters and clients ask the same questions about skills, projects and experience. The assistant answers them from the actual documents instead of a static page.
Recruiters, clients and collaborators who want answers about my work without reading every page.
A recorded walkthrough from architecture to deployment. It loads from YouTube when you press play.
I built the whole system: ingestion, preprocessing, embeddings, the RAG chain, the API, the chat front end and the deployment.
A RAG assistant that answers questions about Haseeb Sagheer’s skills, projects and experience from his own documents.
Yes, at ask.haseebsagheer.com. The code is on GitHub.
OpenAI embeddings and an OpenAI chat model for answers, Pinecone for vector search, and FastAPI for the backend.
From documents in a Google Drive folder, which are ingested automatically when new files appear.