Sagar Penneru

At NVIDIA, I design and ship production AI agents using transformer models and retrieval-augmented generation, with a focus on enterprise evaluation, reliability, and scalable retrieval.

I build end-to-end ML pipelines with Docker, Kubernetes, and MLflow, and optimize semantic search, embeddings, vector databases, and context workflows to improve retrieval accuracy and inference efficiency. I also develop distributed PyTorch and CUDA training workflows, production monitoring, drift detection, regression testing, and automated evaluation systems.

Previously at Dell Technologies, I engineered predictive models combining anomaly detection and classification that improved enterprise dataset accuracy by 25%. I built feature engineering and preprocessing workflows across Python, SQL, and Apache Spark for more than 10 million records.

I'm focused on translating ambiguous product and customer problems into reliable AI systems, combining forward-deployed engineering, full-stack delivery, MLOps, and practical ML judgment.