Vignesh Murali

Projects :

Data Feed Quantity Assurance Automation:

• Developed machine learning model to automate validation and forecasting of data feed quantities across multiple databases.
  Implemented configurable pre-check rules to assess data feeds against business thresholds, reducing need for manual monitoring.
  Used Prophet library to forecast expected data feed counts based on historical trends and country-specific holidays, enhancing predictive accuracy and reducing false alarms.
  Automated real-time alerts for feed anomalies, enabling proactive interventions to ensure data integrity.
  
• KPI Dashboard Extraction and Reporting Automation: Designed and implemented an automated process to extract monthly KPI dashboards from Tableau Server for various countries.
  Automated generation of country-specific Excel reports and distributed them to key stakeholders, improving decision-making and operational transparency.
  Streamlined reporting processes, eliminating manual errors and reducing reporting time by 95%.
  
• Improved data feed quantity accuracy by 20% through automated anomaly detection and timely interventions.
• Reduced manual reporting efforts by 95%, saving over 120 man-hours monthly across departments.
• Scaled automation solutions to handle over 50 countries, enabling seamless expansion and flexibility for future reporting needs.
• Achieved zero discrepancies in reporting, enhancing reliability and accuracy of delivered insights.

Languages: Python (ML model development, automation), SQL

Libraries: Prophet (forecasting), Pandas, openpyxl, scikit-Learn, tableau_api_lib, SQLAlchemy & CX-Oracle

Database : Oracle

Version Control : Git

Alerting Systems: SMTP Server (smtplib)

Projects :

1) Fraud Early Warning System (FEWS)

• Objective: Developed system to detect and flag unusual customer spending patterns and initiate immediate verification processes for high-risk transactions.
• Description:
  Developed machine learning-based system to analyze customer spending behavior, categorizing transactions into risk buckets (low, medium, high) based on factors such as transaction amount, location, and frequency. High-risk transactions were flagged, recorded in separate database, and real-time notification was sent to customer support to initiate call-back verification process within 3 hours of transaction.
  The system used anomaly detection algorithms, incorporating neural networks to predict abnormal behavior patterns.
• Achievements:
• Reduced fraud occurrence by 25% through early identification of suspicious transactions.
  Successfully integrated real-time alerts for customer support, improving fraud response time to under 3 hours.
  Achieved 95% accuracy in correctly flagging suspicious transactions, minimizing false positives.

2) Predictive Analytics for Card Usage

Objective: A predictive model to forecast card usage trends, enabling optimized resource allocation and informed decision-making within the organization.

Key Responsibilities:

• Data Collection and Preparation: Gathering and preprocessing historical card usage data from various sources, ensuring data quality and consistency for analysis.
• Time Series Analysis: To identify seasonal trends and patterns in card usage, helping to understand customer behavior over time.
• Model Development: Create and validate predictive models using statistical and machine learning methods to forecast future card usage trends accurately.
• Actionable Insights: Analyze forecasted trends to generate actionable insights for marketing strategies, helping marketing team tailor promotions to customer behavior.
• Inventory Management: Collaborate with inventory management teams to ensure adequate stock of promotional items based on predicted usage patterns.
• Performance Monitoring: Continuously monitor model performance and update forecasts as new data becomes available, ensuring accuracy and relevance.

Languages: Python (ML model development, automation), SQL & Java

Libraries: Pandas, Time Series Analysis (SARIMAX), Scikit-Learn & CX-Oracle

Database : Oracle

Version Control : SVN

Alerting Systems: SMTP Server & Flask

Project :

Automated Loan Processing System (ALPS)

Overview:
ALPS is an innovative and workflow-driven application designed to automate the booking of home and personal loans for customers at Citibank India. This system was developed to address the growing demand for Mortgage and Ready-cash origination volumes, streamlining the loan processing experience for customers and enhancing operational efficiency.

Key Features:

• Highly configurable and integrated platform that automates various workflows and stages within the loan booking process.
• Enhances customer experience by providing seamless and efficient loan application process.

Roles & Responsibilities:

• Actively participated in development and testing phases, ensuring high-quality deliverables.
• Led production release and implementation activities, ensuring smooth deployment and integration.

Languages: Java (Struts2)

Database: SQL

Application Server: WebLogic 12c

Version Control: SVN

Other Tools: iSignViewer, Putty