Final Year Project

A help desk that answers in the language you asked in

A microservice-based AI chatbot that understands and responds in English, Hausa, Yoruba, and Igbo — built for the university help desk.

See it in four languages

StudentWhen is the school fee deadline?

Help DeskThe deadline for this semester is listed on the student portal under Fees.

StudentYaushe ne ranar ƙarshe na biyan kuɗi?

Help DeskRanar ƙarshe na wannan zangon tana kan portal na ɗalibi a ƙarƙashin Fees.

StudentÌgbà wo ni ọjọ́ ìkẹyìn fún sísan owó ilé-ìwé?

Help DeskỌjọ́ ìkẹyìn fún sémesítà yìí wà lórí portal akẹ́kọ̀ọ́ lábẹ́ Fees.

StudentKedu mgbe ụbọchị ikpeazụ ịkwụ ụgwọ akwụkwọ?

Help DeskỤbọchị ikpeazụ maka semester a dị na portal nwa akwụkwọ n'okpuru Fees.

How it works

A student's message is routed through a chain of small, focused services rather than one monolithic model — which keeps language detection, intent understanding, and translation independently testable and swappable.

01

Streamlit UI

Student submits a question in any of the four supported languages.

02

fastText

Detects which of the four languages the query is written in.

03

DistilBERT

Classifies intent so the query maps to the right knowledge base entry.

04

Knowledge Base

A structured JSON store of help desk answers is queried for a match.

05

deep-translator

Translates the response back into the student's original language.

Python 3.11 Streamlit simpletransformers DistilBERT fastText deep-translator

Project details

Institution

Nile University of Nigeria

Supervisor

Prof. Noah Akande

Languages Supported

English, Hausa, Yoruba, Igbo

Architecture

Microservices, JSON knowledge base

Meet the developer →