Fund UCalgary
An AI-powered funding recommendation system that matches University of Calgary researchers with relevant scholarships, grants, and awards using LLMs, RAG, and vector embeddings.
The Challenge
Researchers spend hours manually searching through hundreds of funding opportunities, often missing relevant grants due to keyword mismatches. Traditional search relies on exact terms rather than understanding research context.
The Solution
Built a semantic search system using ChromaDB for vector storage and Groq's Llama 3.3 70B for intelligent matching. The system crawls researcher profiles, extracts their research areas, and finds funding opportunities using weighted cosine similarity across multiple dimensions.
Key Features
- Automatic researcher profile crawling and data extraction
- Vector search with ChromaDB for semantic matching
- Weighted scoring: 40% area match, 35% bio-summary, 25% eligibility
- LLM-powered reasoning explaining why each opportunity matches
- Shortlisting and award management with PDF parsing via Llama Cloud
Tech Stack
Groq-hosted 70B parameter model generating match explanations and structured evaluations
Vector database storing funding opportunity embeddings for semantic similarity search
Orchestration framework connecting embeddings, prompts, and structured LLM outputs
TypeScript frontend with Vite, Tailwind CSS, and DaisyUI for the researcher interface