AI Agent Evaluation Specialist

Implemented an Agentic AI solution using Smart Agentic SDK to automate deceased verification, replacing manual LexisNexis checks.

Built multi-agent workflows to cross-verify obituary and certificate data.

Designed a 10-point confidence model and automated decision routing, cutting manual review by 75% and boosting accuracy to over 90% through real-time API integration and compliance alignment.

Developed a RAG pipeline with FAISS vector search and AWS Bedrock Claude 3.7 to convert BRD rules into structured JIRA stories.

Integrated React and Flask architecture for seamless automation, improving story generation speed by 70% and consistency by 40%, and standardized backlog creation through LLM-based summarization.

Designed a TOPSIS-based ML model to trigger dynamic credit line increases using behavioral and risk data.

Built propensity models predicting six-month customer engagement with over 90% accuracy, automating credit decisions and improving timing precision by 35%.

Built a RAG-based conversational assistant using LangChain, OpenAI API, and ChromaDB for contextual document Q&A.

Implemented session-aware retrieval chains and deployed the application via Streamlit.