LinkedIn's SLM Framework for Job Understanding

LinkedIn has developed a unified semantic modeling framework for large-scale job understanding, powered by a small language model (SLM). The framework addresses the challenge of transforming unstructured job postings into standardized attributes. It involves fine-tuning an open-source SLM with synthetic tasks augmented by reasoning traces, targeting both taxonomy-guided classification and entity extraction. A multi-adapter architecture with attribute grouping is introduced to enhance scalability and efficiency. The system achieves robust zero-shot generalization across structured and unstructured contexts.