Upendra Golla

Senior Ai Ml Engineer @Citi Bank

Tampa, FL, US

At Citi Bank, I serve as a Senior AI/ML Engineer, leading the design and delivery of enterprise-scale Generative AI and machine learning solutions to enhance customer service and automate knowledge-driven workflows. I architected a scalable Retrieval-Augmented Generation (RAG) platform using LangChain and AWS Bedrock, enabling accurate, citation-backed responses and reducing query resolution time by over 60%.I developed a reusable reference architecture standardizing document ingestion, chunking, embeddings, and semantic retrieval, enabling consistent adoption of AI solutions across multiple teams. I also built advanced multi-agent orchestration systems using LangGraph, where orchestrator agents dynamically route queries to specialized agents for retrieval, summarization, and NL-to-SQL tasks, supporting complex multi-step interactions.To handle large-scale data, I implemented high-performance pipelines using PySpark and AWS Glue, processing over documents with automated quality validation. I integrated Pinecone and FAISS for low-latency semantic search and improved retrieval accuracy through optimized chunking and embedding strategies.I established evaluation frameworks using RAGAS to measure response quality and continuously improve system performance. Additionally, I implemented monitoring using AWS CloudWatch to track latency, hallucination rates, and retrieval precision, reducing issue detection time significantly.I developed scalable API services using FastAPI and Docker, deployed on AWS SageMaker with 99.9% uptime. I automated CI/CD pipelines using GitHub Actions, reducing release cycles from days to hours.I ensured strong governance by implementing RBAC, IAM, prompt versioning, and guardrails aligned with FISMA standards, while collaborating with legal teams to maintain compliance and security across all AI systems.