Hi there!

I’m Elan Markowitz, a researcher passionate about the future of AI. My focus is on combining graphs and large language models, with further expertise and interest in agentic systems, inference-time compute, and AI safety.

I’m currently a final year PhD student at the University of Southern California, working under the guidance of Greg Ver Steeg and Aram Galstyan. I’m open to employment opportunities starting in Summer 2025.

Education

Prior to U.S.C., I studied at UCLA where I received my B.S. in Computer Science (2018).

Areas of expertise

Scalable Graph Learning

• Considerable experience in graph learning
• Trained graph and hybrid language-graph models
• Developed scalable algorithms for graph learning
• Combined LLMs and Graphs in many novel research directions (model editing, retrieval, etc)

Agentic AI

• Developed Agentic AI systems for knowledge graph retrieval and reasoning at Amazon.
• Trained LLM web-agents using synthetic data from tree search systems.

Inference Time Compute/LLM Search

• Early adopter of inference time search.
• Utilized LLM-based inference-time search for agentic applications as early as 2023
• Incorporated inference-time search for knowledge graph retrieval agents at Amazon
• Incorporated LLM search for both inference and synthetic data generation for web agents at MultiOn (now Please.AI)

Distributed Learning

• Experience training large language models on multi-node GPU clusters.

• Wrote custom distributed learning code in order to train large graph models on multiple GPUs (prior to widespread framework support).