SIDDHARTH JAIN

KEY PROJECTS

Customer Effort Score (CES)

• Designed and developed enterprise-scale Customer Effort Score (CES) solutions to measure customer effort across IVR, Voice, Web, and Mobile channels.
• Built scalable PySpark pipelines for processing large volumes of customer interaction data.
• Developed standardized effort-scoring methodologies using interaction duration, transfers, and journey-level metrics.
• Created reusable data products supporting customer experience analytics and business reporting.
• Partnered with stakeholders to translate business requirements into scalable analytical solutions.
• Developed interaction-level and journey-level CES datasets used for downstream analytics and reporting.

Conducted analysis to evaluate relationships between customer effort metrics and payment success outcomes. Tech Stack: Python, PySpark, SQL

• Developed an end-to-end framework for measuring customer effort across multiple service channels.
• Created interaction-level and journey-level effort datasets supporting enterprise analytics.
• Implemented duration standardization methodologies and effort signal engineering techniques.
• Analyzed relationships between customer effort and payment success outcomes.

Impact

• Enabled business teams to identify high-effort customer journeys and prioritize customer experience improvements.
• Delivered standardized CES datasets for downstream reporting and analytics consumption.

Named Entity Recognition (NER) Model in Customer Identification

• Built machine learning models for extraction of authentication-related entities from unstructured customer conversations.
• Collaborated with senior data scientists and engineering teams on model packaging and deployment activities.
• Participated in model validation, testing, deployment readiness reviews, and performance evaluation.
• Supported deployment workflows in cloud-based environments.
• Improved structured information extraction capabilities for downstream analytics and reporting systems.

Tech Stack: Python, PySpark, NLP

• Designed scalable text-processing pipelines for customer interaction analysis.
• Implemented transcript segmentation and context-aware information extraction workflows.
• Built reusable authentication detection logic across multiple interaction channels.

Impact

• Improved automation of authentication analysis workflows.
• Reduced dependency on manual transcript review processes.
• 
  

ASC JOURNy – Customer Defection Prediction

Tech Stack: Python, Machine Learning, Random Forest, Deep Learning

Business Objective: Predict customers likely to defect (churn) at the end of their six-month policy term and identify early defectors to support proactive retention campaigns.

• Developed a customer defection prediction framework supporting customer retention initiatives.
• Built a baseline Random Forest model using the top 30 predictive features identified through feature importance analysis.
• Generated baseline customer risk probabilities ("Knowledge Score") used as the foundational churn risk indicator.
• Developed the JOURNy deep learning model to capture customer lifecycle behavior through event-driven modeling.
• Modeled sequential customer events such as vehicle additions, driver additions, and policy-related activities occurring during the customer lifecycle.
• Combined event history with baseline risk scores to dynamically update customer defection probabilities after each significant event.
• Implemented threshold-based risk identification logic to target high-risk customers before policy expiration.
• Generated weekly lead files used by business teams to execute customer retention campaigns.
• Enabled identification of both end-of-term defectors and early defectors (3–4 months prior to policy expiration).

ML Analytics & Data Science Initiatives

• Conducted statistical analysis and hypothesis testing to evaluate relationships between customer behavior, effort metrics, and business outcomes.
• Developed automated PySpark workflows reducing manual analysis effort and improving reproducibility.
• Worked closely with business stakeholders, data scientists, and engineering teams throughout project lifecycles.

Presented analytical findings and technical recommendations to technical and non-technical audience