FBN.2024.08
EduBot
AI engineer, sole build
Django REST Framework, Llama 3, Vite, TypeScript, 2024
L3
Llama 3 as the reasoning engine
24/7
office hours, in effect
SPA
real-time Vite and TypeScript client
The problem
Students need course-aware help at midnight, not during office hours, and general-purpose chatbots have never read the syllabus.
The approach
Django REST Framework carries the sessions and history; Llama 3, shaped by prompt engineering for teaching, does the reasoning; a Vite and TypeScript single-page interface keeps the conversation fast enough to stay in flow.
The outcome
A contextual course tutor that answers real questions in real time, with sessions and history held securely.
What it does
3 capabilitiesAnswers that know the course
Prompt engineering keeps Llama 3 inside the curriculum instead of drifting into generalities.
Memory across visits
Secure APIs carry sessions and chat history from one night of revision to the next.
Conversation at speed
A single-page interface with the latency a student needs to stay in flow.
The strata
How it standsClient
Vite and TypeScript SPA with a real-time chat interface
API
Django REST Framework: sessions, history, authentication
Reasoning
Llama 3 with teaching-specific prompt engineering
Next exhibit — IX
AI-Driven Customer Insights