Principal Cloud Solution Architect: Advanced Analytics And Ai
Current• Developed the Azure OpenAI Batch Accelerator. The accelerator is a reference implementation for leveraging the Azure OpenAI Batch API to process large sets of files in Azure Data Lake Storage. It provides a flexible configuration, including support for continuous or one-time processing of batch files. The project is designed for extension and testing, rather than production use, and facilitates the automated processing and cleanup of files, metadata generation, and error handling. The tool is ideal for those looking to integrate Azure OpenAI capabilities into batch processing workflows• Led the design and development of a prototype for the automated extraction, categorization, and analysis of medical literature for automated comprehensive literature review. The prototype developed is configurable and uses GPT-4o along with the Azure Document Intelligence service to segment, analyze, and extract data from PDFs, presenting the results to the user in an aggregated structured output. Early estimates project a significant reduction in workload for end-users along with 100s of thousands in terms of cost savings. • Led the design, prototype development, and provided production development guidance of a cloud-native platform for the identification and explanation of transactional anomalies in complex financial data, using machine learning, for a major pharmaceutical company. The project uses gradient-boosted decision trees to automatically identify anomalies and then, along with specific context information, uses GPT-4o to explain those anomalies.• Led the prototype development of a machine learning-based recommendation system for fine-grained sales data. The prototype model demonstrated a deal classification accuracy of over 92% and the project is currently being further developed internally by the customer.
