Implemented a Natural Language based Model Explanations feature for ML & DL Models, leading to a 25% increase in user trust and confidence in AI-driven decisions and outputs, positively impacting user satisfaction and engagement. Empowered end users to effortlessly extract actionable insights from both structured and unstructured data, leveraging cutting-edge Generative AI technology. This accessibility directly contributed to data-driven decision-making, elevating the quality of user insights.Proactively addressed model bias with the Model Fairness feature, reducing bias-related issues, leading to more equitable and unbiased AI outcomes, thus bolstering user trust and ethical standards.Developed chatbot for house insurance claims processing using GPT-3.5-turbo through LLMOps via mlflow, incorporating severity prompt, chat history integration, and context-driven conversation flow, resulting in a 30% reduction in claims processing time and improved customer satisfaction through streamlined and accurate claim assessment.Leveraged unsupervised machine learning to analyze a diabetes dataset, providing valuable insights for optimizing patient treatment strategies and improving diabetes management.Led the design and development of a comprehensive Machine Learning Operations (MLOps) library encompassing 10 key features, including Feature Engineering, Model Registry, Model Explanations, Auto-Deployment, and more.Implemented Automatic Generation of APIs for inference significantly reducing deployment errors and expediting the delivery of AI-driven solutions by 20%, resulting in greater efficiency and reduced operational costs for end users.

• Executed impactful end-to-end machine learning and deep learning solutions, employing MLOps, Docker containerization, and Kubernetes pipeline for seamless scaling, continuous integration, and deployment.• Revolutionized non-invasive CAD testing with an innovative MLROM Model, slashing detection time by 50% while significantly enhancing diagnostic efficiency.• Achieved an exceptional 80% reduction in time to market by harnessing the power of data-driven predictive modeling, utilizing Keras subclassing API, to accurately forecast gas turbine engine performance.• Created an optimized Reduced Order Model for Compressor Surge Detection, outperforming by 3 milliseconds at peak design efficiency.