Principal Staff Engineer, AI Infrastructure

At LinkedIn, our approach to flexible work is centered on trust and optimized for culture, connection, clarity, and the evolving needs of our business. This role may be remote or hybrid. At LinkedIn, hybrid roles are performed both from home and from a LinkedIn office on select days, as determined by the business needs of the team. Remote roles are performed from the designated home work location upon time of hire, and any changes to this home work location requires a review of remote status and approval. 

We’re hiring a Principal Staff Software Engineer to lead LinkedIn’s GPU-Based Retrieval Platform, a foundational AI infrastructure stack that powers candidate generation and retrieval across Feed, Ads, Search, Talent, and other critical product experiences. The platform sits on the hot path of billions of member interactions each day, making this one of the highest-leverage technical leadership roles within LinkedIn’s AI Infrastructure organization.

In this role, you will own the platform’s technical direction and architecture end to end, spanning large-scale indexing and retrieval, low-latency distributed serving, GPU scheduling, memory efficiency, batching, and kernel-level optimization. You will drive improvements in throughput, tail latency, retrieval quality, reliability, and cost, directly influencing member engagement, business outcomes, and AI engineer productivity across the company.

The GPU-Based Retrieval Platform team works at the intersection of GPU systems, distributed serving, information retrieval, machine learning, and product engineering. You will partner closely with teams across Feed, Ads, Search, Talent, Modeling, and Infrastructure, while setting the technical direction for a multi-team platform, influencing cross-company architecture, and mentoring senior engineers.

Responsibilities:

• Set the long-term technical strategy and architecture for LinkedIn’s GPU-Based Retrieval Platform.

• Lead the design and evolution of large-scale indexing, candidate generation, vector search, and retrieval-serving systems.

• Optimize GPU performance across CUDA, Triton, memory hierarchies, batching, scheduling, and multi-GPU communication.

• Improve throughput, QPS per GPU, tail latency, recall quality, reliability, and infrastructure cost.

• Build scalable, observable, and highly available multi-tenant serving systems for high-QPS production workloads.

• Make critical architectural trade-offs across latency, quality, capacity, model complexity, and cost.

• Partner with Feed, Ads, Search, Talent, Modeling, and Infrastructure teams to shape the platform roadmap.

• Evaluate emerging GPU technologies, retrieval architectures, and serving frameworks for adoption at LinkedIn.

• Lead complex cross-organizational initiatives from architecture and design through production rollout and adoption.

• Mentor senior engineers, raise the technical bar, and influence AI infrastructure strategy across LinkedIn.