At Infor, I designed and deployed a GenAI-powered API Policy Generation System using LLMs and RAG. It generated context-aware policies by retrieving information from enterprise documentation through embeddings and vector databases.
I integrated LLMs with enterprise systems using MCP, AWS Bedrock, and Lambdas. My Python microservices used function-calling workflows and REST APIs to retrieve API metadata and backend data, and handled 2 million API requests per day across 200+ enterprise clients.
For my wind speed forecasting project, I researched how outliers affect LSTM models and developed an outlier treatment methodology. It reduced MSE by 95% and increased R² by 10%.
I also developed a U-Net model for MRI image segmentation and published work on wind speed data outlier detection and imputation at the IEEE 4th International Conference on Sustainable Energy and Future Electric Transportation (SEFET), where it received an award.
