Vahdettin Karataş
Data & ML engineering — retrieval systems, APIs, deployment
  • Location:
    Prague, Czech Republic
Technical focus
  • RAG & document QA
  • Embeddings & vector search
  • FastAPI & OpenAPI
  • Docker & VPS deployment
  • Reproducible indexing pipelines

Interactive Q&A demo

RAG Document QA — browser demo against the live /ask endpoint.

Pick an example or type a question against the eight-document policy corpus. Answers and cited sources come from the live API — useful when reviewers want evidence, not slides.

  • 8policy documents
  • /asklive API
  • 3fields per source

How it works

  1. Embed the question — same model contract as the indexed corpus.
  2. Retrieve top-k — FAISS returns the closest policy/runbook chunks.
  3. Answer with citations — the API returns answer plus sources[] (document, page, chunk_id).

Demo corpus is the fixed sample set in data/raw_docs/; production work scopes your documents, auth, and hosting boundary.

Ask the demo corpus

Fixed index: 8 synthetic policy documents (SLA, billing, refunds, incident response, support workflow, onboarding, pricing, internal FAQ) from data/raw_docs/. Example chips below map to those themes.

Calls POST /ask on the deployed API from your browser. If the request is blocked (CORS or API key on the server), use API docs or curl instead.

Try an example (fills the box; then press Ask):

RAG Document QA

Portfolio artifact · live API on VPS

© Vahdettin Karataş. All rights reserved.