Vivek Chaurasiya is a Staff Software Engineer with over 12 years of experience building large-scale distributed systems, recommendation platforms, and AI infrastructure. He has led high-impact engineering initiatives at Snap Inc., Microsoft, and OfferUp, specializing in graph infrastructure, real-time recommendation systems, embeddings, search, and cloud-native platforms. His work has driven significant improvements in user engagement, developer productivity, and infrastructure efficiency, including multi-million-dollar annual cost savings. Vivek holds an M.S. in Computer Science from Purdue University and a B.Tech. from IIT Roorkee, with expertise in Java, Python, Golang, AWS/GCP, Elasticsearch, GraphDBs, VectorDBs, and Kubernetes

Embedding Based Retrieval in Friend Recommendation Friend recommendation systems like Friends-of-Friends (FoF) boost user engagement but struggle with cold-start scenarios, real-time latency beyond 2 hops, and capturing graph topology. To overcome this, we deployed a novel Embedding Based Retrieval (EBR) system using Graph Neural Networks at Snapchat. Supporting billions of users, our EBR system complements FoF by enabling retrieval beyond 2 hops and providing natural candidate ranking. Online A/B testing demonstrated statistically significant increases in friendships created across both low- and high-density markets