Churn Prediction
Churn modelling across 3, 7, 14 and 30-day horizons, with behavioural signals used to support retention analysis and live campaign workflows.
Behaviour → Time windows → Churn risk → Retention
WORK & CASE STUDIESProblems with
Problems with
real context.
LIVE SYSTEM / 01DECISION NETWORK
DATA
RISK
AI
SYSTEM
DECISION
MOVE POINTER TO STRESS THE MODEL84.7
LIVE SYSTEM / 01DECISION NETWORK
DATA
RISK
AI
SYSTEM
DECISION
MOVE POINTER TO STRESS THE MODEL84.7
Professional projects across banking, AI, analytics and business analysis. Confidential data and internal code stay private; the thinking and contribution can still be shown.
Business Analyst Intern · Asset Management & International Operations
Overdraft-account products, requirements, user stories, process flows and collaboration across analysis, solution design and development.
2026—Now
Credit Default / PD Model
Built an complete probability of default model from real-world banking data and presented methodology, results and business implications to Risk Management professionals.
2026
AI Identity & Document Verification Model
AI model for reading identity and official-document information, comparing extracted fields with structured records, and detecting inconsistencies or mismatches.
2025
Churn, Growth & Marketplace Intelligence
Churn prediction, driver-facing AI work and performance analysis across marketplace, operational and product metrics.
2025—26
Software Development · Core Banking
Enterprise banking applications using .NET, Java and COBOL, including core banking and international money-transfer processes.
2024
PROJECT SCREENSHOTS / PRESENTATION MEDIA. TO BE ADDED
Presentation.
A recurring part of my work is translating technical or analytical depth into a story that stakeholders can use.
COMMUNICATION
Technical → clear → actionable.
Model presentations · business storytelling · data visualisation · stakeholder communication · technical-to-business translation.
MARTI TECHNOLOGIES / DATA & AI
Identity & Record Consistency
AI-assisted reading of identity and official-record information, comparison with structured data, and cross-field consistency checks to surface mismatches.
Read → Extract → Match → Check → Flag