AI Tutoring Platform Customization

Job ID: 39831737

Budget: $2 – $8 USD

Project Overview:

We are building a modular AI tutoring platform for Undergrad students. I have a fork of OpenWebUI that needs customization to support:

Module-Based Data Loading – The AI should only load embeddings relevant to the requested module (Math, Science, etc.) instead of all modules at once.

Multi-Model Support – Integrate three LLM tiers: Lite, Regular, Heavy. The system should select the appropriate model based on mode or prompt.

Dynamic Retrieval System – Implement on-the-fly retrieval from vector stores for each module, including caching to reduce repeated queries.

Session Management – Maintain short-term context per session without storing any sensitive student data (COPPA compliance).

Prompt / Mode Switching – Suggested prompts should act as mode switches (Tutor, Exam, Socratic, etc.).

Requirements:

Experience with Python, FastAPI/Django, or similar backend frameworks.

Experience integrating LLMs and vector databases (e.g., FAISS, Milvus, Weaviate).

Understanding of multi-model orchestration and prompt engineering.

Ability to write clean, documented, modular code.

Optional: experience with OpenWebUI or similar UI for LLM frontends.

Deliverables:

Customized OpenWebUI fork with module-based loading fully implemented.

Multi-model selection logic integrated.

Retrieval + caching system per module.

Documentation for setup and module integration.

Optional: guidance on connecting front-end queries to the backend.