At Epsilon, I work with business stakeholders to turn ambiguous requirements into production AI and agentic solutions. I deploy customer-facing workflows using Python, FastAPI, LangChain, and LangGraph, and evaluate RAG pipelines with RAGAS, reaching approximately 80–88% across evaluation metrics.
I previously developed RAG systems at Maybank, where document retrieval and question-answering achieved 85–90% answer accuracy. I also built data pipelines and deployed machine learning models, including REST APIs with average inference latency under approximately 300 ms.
At McKesson, I developed Python and FastAPI applications, ETL pipelines, and machine learning workflows. I also built AgentForge, a stateful multi-agent software engineering runtime with coordinated development workflows and an automated evaluation harness.
