LinkedIn builds AI-powered semantic job search using LLMs, embedding-based retrieval, and GPU infrastructure
Traditional keyword-based job search failed to capture nuanced user intent and required exact keyword matches, while LinkedIn's legacy AI pipeline had grown into a multi-stage system duplicated across many channels that made relevance improvements difficult.
How it works
Common implementation structure
How this type of workflow is generally built, generalized across documented cases — not tied to any one vendor's stack. Click any stage to read what happens there. Specific products that implement these stages appear in “Tools commonly seen” below.
Stage 1 · Natural language query submitted
Members describe the job they want in their own words rather than supplying exact keyword queries.
Tools used
LLMembedding-based retrievalGPUvector databaseRAGTool CallingFSDPDeep and Cross Network architecture
Outcome
LinkedIn delivered a semantic job search system that understands natural language queries, reduces pipeline complexity by an order of magnitude, processes annotation grades at scales well beyond human capacity, and serves retrieval results in a few milliseconds per query, while significantly enhancing developer velocity.
What failed first
LinkedIn's legacy candidate selection and ranking models used fixed taxonomy-based methods and older LLM technology that lacked capacity for deep semantic understanding, and the multi-stage pipeline had become too complex and duplicated to permit reliable relevance improvements.
Results
Time saveda few milliseconds per query
Volumereducing the model pipeline complexity by an order of magnitude