Enterprise Graph Retrieval Intern

About the team

The team builds Agentic Retrieval systems that power intelligent search, retrieval, and reasoning across Zoom’s enterprise platforms. We work at the intersection of knowledge graphs, retrieval systems, large language models, and production engineering to enable scalable, reliable access to enterprise knowledge. Interns are treated as full contributors and work on real problems alongside experienced engineers and scientists.

What you can expect

As a Machine Learning, Applied Scientist, or Research Engineer Intern, you will contribute to Enterprise Graph Retrieval within the Agentic Retrieval systems that support Zoom. You will work on building graph-based retrieval and reasoning capabilities that enhance enterprise RAG systems and enable intelligent AI agents. Your work will focus on designing, building, and evaluating graph-powered retrieval systems that operate on real enterprise data and directly impact users across Zoom products. You will collaborate closely with engineers and product partners, contribute to technical discussions, and deliver working components, evaluations, and demos that showcase graph-enhanced search and agentic reasoning.

Responsibilities:

Depending on project needs and individual strengths, you may work in one or more of the following areas:

Design and model enterprise graph schemas

• Design flexible graph schemas representing enterprise relationships, including:

  • Document-to-document relationships such as references, versions, and derivatives

  • Meeting-to-content relationships such as notes, presentations, and follow-ups

  • User-to-content relationships such as authorship, editing, and viewership

  • Topic-based relationships across enterprise content

• Produce clear schema documentation and entity relationship models

Build graph construction pipelines

• Build data processing workflows to extract relationship signals from raw enterprise content

• Apply LLM-based extraction techniques combined with enterprise knowledge bases for entity resolution

• Transform extracted signals into graph edges and properties

• Load, update, and maintain graph data in a graph database such as Amazon Neptune

• Optionally ensure consistency with enterprise access control and permission models

Implement graph-aware querying and retrieval

• Implement graph-aware retrieval mechanisms

• Translate natural language queries into graph queries

• Apply LLM-based query splitting and triplet extraction

• Perform schema-driven graph traversal, subgraph extraction, and multi-hop retrieval

• Support contextual and relationship-based retrieval use cases

Integrate and evaluate retrieval systems

• Integrate graph-based retrieval signals into a unified enterprise retrieval pipeline

• Design evaluation metrics to measure retrieval quality improvements

• Analyze relevance, latency, and scalability trade-offs

• Build end-to-end demos showcasing graph-enhanced search and agentic reasoning

What we're looking for:

(Required)

• Currently pursuing a BS, MS, or PhD in Computer Science, Machine Learning, AI, or a related field

• Demonstrate strong programming skills in Python; Java is a plus

• Apply solid foundations in algorithms, data structures, and system design

• Show interest in information retrieval, RAG systems, or knowledge-centric AI

Preferred skills

Graph & Knowledge Representation

• Experience with knowledge graphs, property graphs, or RDF-based systems

• Familiarity with graph query languages (Gremlin, SPARQL, Cypher, or similar)

• Understanding of graph modeling, schema design, and relationship semantics

• Experience with graph databases (e.g., Amazon Neptune, Neo4j, JanusGraph)

GraphRAG & Retrieval

• Hands-on experience with GraphRAG or hybrid graph + vector retrieval systems

• Knowledge of combining symbolic graph reasoning with neural retrieval

• Experience integrating graph signals into ranking or retrieval pipelines

• Familiarity with subgraph extraction, path-based reasoning, or multi-hop retrieva