node classification, link prediction, embeddings, and Graph Neural Networks. • Design GraphRAG and knowledge-grounded security assistants that combine graph traversal, vector retrieval, structured evidence, LLM reasoning, citations, and human approval controls. • Partner with SOC, threat intelligence, IAM, vulnerability management, cloud security, architecture, data engineering, and product teams to convert operational … Practical use of MITRE ATT&CK, STIX/TAXII, CVE, CWE, CAPEC, NIST frameworks, CIS Controls, and common threat-intelligence vocabularies. • GenAI and GraphRAG : LLM-based extraction, retrieval orchestration, agent/tool integration, prompt design, evaluation, grounding, guardrails, explainability, and evidence traceability. • MLOps and observability: Experiment tracking, model versioning, deployment ...