Project: Prajna Framework

From TMTS LLC (July 2018 - May 2019) [cite: 275, 279]

Business Problem Statement

The need was to build an end-to-end Machine Learning Life Cycle for internal applications at TMTS LLC, focusing on understanding employee mindset, behavior analysis, and work-life balance, and to introduce AI-powered chatbots. Ensuring quality assurance and diminishing risks in ML project success was crucial. [cite: 280, 283, 285, 286]

Solution

• Built the End to End Machine Learning Life Cycle for Prajna and deployed it to SDLC. [cite: 280]
• Identified use cases for Machine Learning in existing internal company applications. [cite: 282]
• Built the Prajna Framework for real-time understanding of employee mindset, behavior analysis, and work-life balance statistics. [cite: 283]
• Reported analysis to Top Level Management on how models learned and how changes occurred in the behavioral analysis of Employees. [cite: 284]
• Introduced Enterprise and Consumer Chatbots, empowered with AI and Automation. [cite: 285]
• Followed CRISP-ML(Q) as a systematic process model for machine learning software development, emphasizing quality assurance to diminish risks. [cite: 286]
• Worked closely with business stakeholders and engineering team to encourage statistical best practices in experimental design, data capture, and data analysis. [cite: 288]
• Troubleshot and found the best prediction Machine Learning algorithms for business use cases, providing high accuracy models. [cite: 289]
• Built a preliminary version of Machine Learning Models for A/B test content for clear decisions related to products. [cite: 290]
• Checked different model performances on various business use cases and generated reports based on metrics, tuning parameters, accuracy, and loss. [cite: 291]
• Retrained ML systems and models as necessary and re-ranked them by their success probability based on reports. [cite: 292]
• Integrated reports into Prajna Frameworks with real-time data visualizations. [cite: 293]
• Contribution led to a 4% increase in sales and model building capacity for developers and prediction performance. [cite: 294]

Tools & Frameworks Used

• Prajna Framework [cite: 283, 288, 293]
• Machine Learning Life Cycle [cite: 280]
• AI and Automation for Chatbots [cite: 285]
• CRISP-ML(Q) methodology [cite: 286]
• Python [cite: 295, 338]
• NumPy, scikit-learn, pandas, matplotlib, TensorFlow, PyTorch [cite: 295]
• APIs (for data analysis) [cite: 288]
• Real-time data visualizations [cite: 293, 356]