About the Role
Adobe is seeking a passionate engineer to join the Adobe Genuine Engineering team. This group protects Adobe's ecosystem from fraud, abuse, and misuse using intelligent systems worldwide.
In this role, you will compose and develop our unified graph along with the detection systems running on it. You will merge multiple isolated fraud graphs into one scalable source that identifies non-genuine and abusive signals across hundreds of millions of users. You will manage the entire lifecycle: starting from raw behavioral and account data, progressing through large-scale graph modeling and ingestion using Databricks and Spark, applying Graph Data Science algorithms and graph ML, and delivering production-ready detection for enforcement.
This role is for an engineer eager to manage graph systems entirely, covering schema, pipelines, community detection, and graph ML rather than using pre-built solutions.
Key Responsibilities
Build and evolve a unified graph schema that merges multiple fraud and account data sources into one consistent, rebuildable model.
Develop and enhance large-scale graph ingestion and feature pipelines on Databricks and Spark, converting raw behavioral and account events into refined graph nodes, edges, and properties.
Apply Graph Data Science (GDS) algorithms including community detection (WCC, Louvain, Label Propagation), centrality (PageRank), and node embeddings (FastRP, Node2Vec) to identify abuse rings, shared-entity clusters, and coordinated fraud.
Develop graph machine learning models, including Graph Neural Networks (GNNs), to extract high-value risk signals from network structure, and incorporate them into downstream ML workflows.
Translate prototypes into production graph systems that are scalable, reliable, and observable, and drive query and inference performance through modeling and serving-side optimization.
Own operational health of the graph platform: incremental refresh, supernode handling, monitoring, and cost efficienc
📌 Software Development Engineer 3 (Noida)
🏢 Adobe
📍 Noida