SLIIT Academic Chatbot Platform
An independently developed academic AI platform that grounds each assistant in module resources and combines document ingestion, vector retrieval and student memory for personalized learning support.
The problem
A university-wide assistant must keep answers within the correct module material, personalize support without losing context and scale beyond a small pilot group.
The approach
I designed module-specific assistants backed by automated document extraction, chunking, embeddings and vector search, then added separate conversation, semantic student and knowledge-base memory layers.
Separated module knowledge so retrieval remains relevant to each learner's academic context.
Automated resource ingestion from text extraction through chunking and embedding generation.
Designed scalable backend services and database architecture for expansion across faculties.
Results
Platform architecture prepared for a planned 2,000+ first- and second-year student rollout.
Personalized learning context across conversation, student memory and module knowledge.
A reusable ingestion pipeline for academic resources and future faculty expansion.
Explore more AI systems and product engineering case studies.