AI/ML Engineer State Street
Architected an Agentic AI platform using Python and LangGraph with more than 12 autonomous agents, reducing manual analyst effort by 40% across investment research and portfolio analytics. Built a GraphRAG knowledge retrieval platform with Neo4j and LlamaIndex across more than 500,000 financial documents, improving retrieval relevance by 32%.
Fine-tuned domain-specific LLMs using PEFT and LoRA, improving response accuracy by 22% while reducing hallucinations by 30%. Developed evaluation and monitoring with LangSmith and RAGAS, predictive risk models using XGBoost and PyTorch, and AML and trade-surveillance pipelines with PySpark and Databricks.
Implemented MLOps and LLMOps workflows with MLflow and GitHub Actions, deployed inference services with Docker and Kubernetes, and introduced governance controls including audit logging, RBAC, data lineage, and Responsible AI practices. Collaborated with portfolio managers, quantitative analysts, compliance teams, and engineers to deliver production AI solutions.