Client project via AfriDev (Upwork agency) · private repository
Aquinas: Scholastic AI Tutor
An end-to-end RAG tutoring platform grounded in Catholic scholastic philosophy, with answer quality measured before release.
- Role
- Backend & AI Engineer
- Status
- Status: Delivered
- Code
- Private client repository
- Stack
- Next.js
- Django REST Framework
- LlamaIndex
- Celery
- Redis
- Docker
- LLM-as-judge
The problem
The client needed a tutor whose answers stay grounded in scholastic philosophy sources instead of an LLM's general knowledge, delivered as a production-ready platform under a strict deadline.
What I built
A Retrieval-Augmented Generation (RAG) pipeline with LlamaIndex over the source material
A Next.js frontend on a modular Django REST Framework backend
Asynchronous background processing with Celery and a Redis broker, plus Redis caching for queries and responses
An LLM-as-judge evaluation pipeline to score answer quality before release
Every service Dockerized for production deployment
Architecture
Request path
- Learner (Next.js)
- Django REST API
- Redis cache
- Celery worker
- LlamaIndex retrieval
- LLM answer
Evaluation loop (AI evaluation / decision step)
- Test questions
- RAG pipeline
- LLM judge
- Quality scores
Green marks an AI evaluation / decision step.
Engineering highlights
- Heavy work runs in background jobs, so the API stays responsive
- Caching repeated queries cuts latency and LLM cost
- Quality is measured with an LLM judge instead of being eyeballed
Contact
Have something to build?
Tell me what you want automated or which AI feature you need, and I'll reply with a concrete plan.
Addis Ababa, Ethiopia (UTC+3). Full overlap with Europe, morning overlap with the US.