What to Build

The 2024 Neodata Hackatania 2 focuses on AI strategies for knowledge management, specifically using graph RAG and agentic systems. Word embeddings, search algorithms, vector databases, and frameworks like LangChain and LlamaIndex have created a huge opportunity for improving decision-making through better knowledge management. 

What to build? Apply graph RAG or similar systems to an enterprise-level problem, such as how to query a large knowledge base while doing time-sensitive work, like hospital triage or emergency services. One excellent source of data you could use is https://ec.europa.eu/eurostat/data/database. Maybe try to make it easy to query the hundreds of databases available from the European government.

Or start with a RAG-based chatbot for enterprise use. Optimize your strategy for a specific use case, innovate around how to improve the quality of RAG retrieval, prompt engineering, best LLMs to use, and more. Or focus on multimodal and surface important information found in diagrams and charts, not just text. Or consider how to fine-tune an LLM with numerical data and summarize everything in one place for the user... 

We are open to a wide variety of submissions. The important thing is that you leverage Graph RAG technology to solve a knowledge management problem.

What to Submit

In order to judge the submission, the jury needs access to your source code (email a Github repository link to info@neodatagroup.ai), and you need to present your work briefly to the public.