360Brew: LLM-based Personalized Ranking and Recommendation — Hamed Firooz and Maziar Sanjabi, LinkedIn AI
AI Engineer World's Fair 2025 · 22:00
← All organizationsLinkedIn
www.linkedin.com ↗
Professional networking, recruiting, and learning
LinkedIn operates a professional network where people find jobs, build connections, share industry knowledge, and learn skills. Job seekers can signal availability through Open To Work, while employers post vacancies and recruiters use AI tools for hiring. LinkedIn Learning offers courses spanning business, software development, and artificial intelligence. Its Hiring Assistant is an AI agent designed to handle repetitive recruiting tasks.
Founded by Reid Hoffman, Allen Blue, Jean-Luc Vaillant, Eric Ly, and Konstantin Guericke, LinkedIn launched in 2003; Dan Shapero is its current CEO. Its engineering contributions include Kafka, a distributed message queue developed internally and open-sourced in 2011. Built to transport large volumes of website activity events with low overhead and horizontal scaling, Kafka supported both real-time services and offline analytics through the same infrastructure.
Microsoft completed its acquisition of LinkedIn in 2016 for a total purchase price of $27.0 billion. LinkedIn’s business includes marketing solutions and Premium alongside its hiring products. The company reported $19.8 billion in FY2026 annual revenue and, in July 2026, recruiters at more than 20,000 companies using its AI-powered hiring solutions. Its company statistics page listed more than 1.3 billion members as of August 2026.
2 talks
Newest first
3 speakers at AIE
Affiliations reflect their AIE appearances, not necessarily current employment.
Start here
- 360Brew: LLM-based Personalized Ranking and Recommendation — Hamed Firooz and Maziar Sanjabi, LinkedIn AI
Start here to learn about experiments with Mixtral and longer user-history contexts, plus audience questions about generalization and multi-item scoring.
Hamed Firooz · Maziar SanjabiAI Engineer World's Fair 2025
- Lessons from Building LinkedIn's GenAI Platform
Use this talk to understand how interchangeable model backends, responsible AI governance, and complementary team skills fit into building a generative AI platform.
Xiaofeng WangAI Engineer Summit 2025
Messages from the stage
Serving large personalization models
The 360Brew presentation connects Brew-XL training and alignment with deployment requirements, including pruning and quantization for high-throughput, low-latency serving.
Shared infrastructure for agent applications
Xiaofeng Wang describes a LangChain-based Python SDK integrated with LinkedIn's Java infrastructure, supported by prompt versioning, conversational memory, distributed agent orchestration, a centralized API skill registry, and OTEL-based observability.
Affiliations reflect each recorded session, not necessarily current employment.