Maharshi Jha and Aygun Aydin, Meta
Privacy compliance demands granular, real-time tracking of data flows at scale. This talk presents the architecture behind Meta's in-memory lineage graph, processing billions of edges across web, warehouse, and AI systems. We cover compressed graph storage, efficient traversal algorithms, and cross-platform data flow mapping. Beyond compliance, we explore how the same architecture enables agentic analysis through interactive graph traversals. The presentation shares practical solutions and architectural lessons from operating at billion-edge scale daily.
Maharshi Jha is a Software Engineer on Meta's Graph Observability team, specializing in privacy-aware data lineage. He architects in-memory lineage graphs processing billions of edges across AI, warehouse, and web systems, enabling privacy compliance and agentic analysis at scale.
Aygun Aydin is an Engineering Manager in Privacy Infrastructure at Meta, specializing in asset understanding and data lineage. He has helped build Meta's data understanding and data flow evaluation frameworks. He holds a masters degree in Software Engineering and brings deep experience in large-scale systems and high-performing teams.
author = {Maharshi Jha and Aygun Aydin},
title = {Architecting Scalable Data Lineage Graph for Privacy Compliance and Agentic Analysis},
year = {2026},
address = {Santa Clara, CA},
publisher = {USENIX Association},
month = jun
}