> Markdown version of [/@gal-shubeli](https://www.wearedevelopers.com/@gal-shubeli). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Gal Shubeli
Builds graph-powered retrieval systems at FalkorDB to improve LLM multi-hop reasoning.

## About

Gal Shubeli builds graph-powered retrieval systems as an AI engineer at FalkorDB. He focuses on modeling knowledge as interconnected nodes, helping large language models reason across multiple hops with higher accuracy.

His work aims to fix AI hallucinations by replacing flat vector retrieval with structured, traversable contexts. In his talk on [scaling GraphRAG](/videos/100025-scaling-graphrag-efficient-knowledge-retrieval-for-ai), he breaks down how this architecture handles complex multi-hop inference.

Before focusing on NLP, he developed machine learning systems for medical ultrasound and earned an electrical engineering master's from Ben-Gurion University.



## Past Sessions

### World Congress 2026 Europe · July 8, 2026
Berlin, Germany
- [Scaling GraphRAG: Efficient Knowledge Retrieval for AI](https://www.wearedevelopers.com/videos/100025-scaling-graphrag-efficient-knowledge-retrieval-for-ai) · 30 min · 🎥 Watch recording



## Videos

- [Scaling GraphRAG: Efficient Knowledge Retrieval for AI](https://www.wearedevelopers.com/videos/100025-scaling-graphrag-efficient-knowledge-retrieval-for-ai) · Gal Shubeli


## Links

- [LinkedIn](https://www.linkedin.com/in/gal-shubeli)