Multilingual Document QA Chatbot
A retrieval-augmented assistant answering questions over a document set in the user's own language — deliberately contained in scope, led and delivered end to end by the team.
- Role
- Team lead & delivery
- Languages
- Multilingual
- Architecture
- Grounded RAG
The problem
A body of documents that people needed answers out of, not a reading list. The scope was narrower than a general assistant — question in, grounded answer out — but the language requirement was not negotiable: it had to work in the languages its users actually write in, not work in English and be translated at the edges.
That constraint does more than it looks like it does. Retrieval quality in morphologically rich, comparatively lower-resource languages is not what it is in English, and the gap shows up exactly where you would rather it did not — in whether the right passage comes back at all.
The approach
I led the team and owned the core logic through to delivery and deployment. The system is deliberately simple, and the simplicity is the decision:
- Grounded retrieval over the document set, so answers come from the source material rather than from model recall.
- Multilingual retrieval quality as the main engineering effort, because that is where this system fails if it fails at all.
- Scope held down on purpose — where the lineage chatbot needed live updates and traceability, this one needed to be dependable and finished.
- Delivered and deployed by the team, rather than handed over as something that worked on someone’s machine.
The outcome
A working assistant, in production, that answers in the language its users actually speak — and one that stayed small enough to stay maintainable.