Kuthuru Sudhir Kumar

About Kuthuru Sudhir Kumar

I am a Senior Generative AI / Machine Learning Engineer with 10+ years of experience designing, building, and deploying large-scale AI, ML, and enterprise Generative AI systems across finance, healthcare, retail, and telecommunications domains.My expertise spans LLMs, GenAI platforms, and end-to-end MLOps, with hands-on experience working on production-grade systems used by millions of users, including large virtual assistants and intelligent automation platforms. I specialize in LLM fine-tuning, Retrieval-Augmented Generation (RAG), prompt engineering frameworks, and multi-modal AI solutions involving text, vision, and speech.I have deep experience working with GPT-4, Claude, Gemini, LLaMA, Mistral, and Hugging Face models, and have built scalable AI systems using LangChain, FAISS, Pinecone, Weaviate, and enterprise knowledge bases. My work focuses on delivering secure, compliant, and high-availability AI solutions on cloud platforms such as AWS, Azure, and Google Cloud, using Kubernetes, Docker, Terraform, and CI/CD pipelines.Beyond technical delivery, I actively collaborate with product, compliance, and business teams to align AI initiatives with measurable outcomes. I also mentor junior engineers and contribute to AI governance, model evaluation, and ethical AI adoption, ensuring solutions are reliable, explainable, and production-ready.

Experience

1. Senior Gen Ai Engineer

   American Express

   Feb 2025 — Present · New York, NY, US

   Working on enterprise Generative AI solutions using GPT-4, Claude, Gemini, and LLMs to enhance large-scale conversational and automation platforms.Implemented multi-stage RAG pipelines (chunking → embedding → retrieval → reranking → summarization) to improve answer relevance in enterprise knowledge retrieval systems.Developed Python-based full stack AI applications using FastAPI backend services integrated with React-based frontends.Built end-to-end GenAI applications using LangChain and Retrieval-Augmented Generation (RAG) to enable enterprise knowledge retrieval.Implemented scalable RESTful APIs in Python supporting AI model inference and real-time data processing.Designed full-stack AI workflows integrating LLM orchestration, vector databases, and web applications.Built enterprise AI platforms combining Python backend, vector search, and AI orchestration frameworks.Developed multi-agent collaboration frameworks where specialized agents (retrieval agent, compliance agent, summarization agent) worked together to deliver accurate financial guidance.Implemented supervisor-agent architecture to manage task delegation and enforce business rules across Erica’s conversational workflows.Designed and implemented Retrieval-Augmented Generation (RAG) pipelines using LangChain, FAISS, Pinecone, and vector databases for accurate, knowledge-grounded responses.Built and standardized prompt engineering frameworks (Chain-of-Thought, ReAct, multi-turn dialogue) to improve response quality and task completion.Integrated Azure OpenAI, AWS Bedrock, and Vertex AI into secure, scalable applications built with Python, Java, and Kotlin microservices.Deployed containerized GenAI services on Kubernetes (EKS/GKE) ensuring scalability, reliability, and high availability.Implemented LLMOps and MLOps pipelines using MLflow, Kubeflow, CI/CD, with monitoring via Prometheus, Grafana, and Evidently AI.

Education

• Webster University

  ITM, Information Technology Management