Implemented a system for automatic information extraction from bill documents, achieving a 95% accuracy rate and reducing manual effort while enhancing data processing efficiency by 70%.Strategically designed data cleaning and preprocessing pipelines for internal signature data, resulting in a remarkable 95% enhancement in data integrity and accuracy.Developed a model which could predict and classify employee attrition using sentiment analysis, yielding a 93% accuracy, enhancing user experience, and bolstering reasons for attrition as part of POC.Developed, maintained, and optimized Python packages for distribution via the Python Package Index (PyPI) using pip, ensuring seamless installation and integration across various environments.

Worked extensively in building, maintaining, and upgrading Anaconda's conda packages for various architectures, primarily utilized in machine learning and deep learning frameworks. These packages were developed using Python, Bash scripting, YAML, and Jinja script. Additionally, also have resolved issues encountered by the open-source community during the installation or development of conda packages.Developed customized Docker images to include specific requirements to use latest RHEL versions with and without Nvidia’s CUDA support for different architecture.Developed Water Quality Assessment project as part of IBM’s Call-4-Code which could predict whether water is potable or not and was developed using IBM’s Watson services like Watson Assistant, Watson Studio, and Watson Machine Learning.Developed Banking Customer Care Assistant Chat-Bot to help users resolve queries related to banking operations. It was developed using IBM’s Watson AI services like Watson Assistant, Watson Studio and Watson Discovery.