// projects

Selected
Work

End-to-end machine learning, analytics, and AI projects built with real data — from raw inputs to deployed models and dashboards.

Churn
02
MLRisk
Customer Churn Prediction & Retention Strategy
K-Means segmentation on 11,260 customers. SVM classifier with SMOTE achieving 88% recall and ROC-AUC 0.915. Retention strategy projected to reduce churn by 15–20%.
Travel Booking Bot
03
NLPAzure AI
Conversational Language Understanding — Travel Bot
CLU model with 3 intents and 3 entities built in Azure Language Studio. Evaluated intent accuracy, entity extraction, and failure modes. Python query script included.
Object Detection
04
Computer VisionAzure AI
Object Detection Pipeline — Azure Computer Vision
Full object detection pipeline using Azure CV API — bounding boxes, confidence scoring, CSV/JSON export. Robustness evaluated across occlusion, small objects, and low-contrast scenes.
Insurance
05
TableauRisk
Insurance Claims Risk & Policy Optimisation
Multi-dashboard Tableau Story on $11M+ in claims across vehicle type, demographics, and travel exposure. Custom KPIs and 6 strategic recommendations on premium recalibration.
SQL
06
SQLBI
E-Commerce Business Intelligence via SQL
10 business questions answered with multi-table SQL — segmentation, inventory, logistics, market basket analysis. Identified $2.5M+ high-concentration inventory and designed discounting logic.
Election
07
MLClassification
Election Vote Prediction
Voting behaviour predicted using Naive Bayes, KNN, and ensemble methods on survey data. Descriptive analysis of candidate speeches included.
Wine
08
Time SeriesForecasting
Wine Sales Forecasting
Multiple forecast models for two wine products — recommended optimum model to predict 12 months of monthly sales with confidence intervals for ABC Estate Wines.
Car
09
TableauRisk
Car Insurance Claim Analysis
Interactive Tableau dashboards exploring claims data — trends, patterns, and actionable recommendations for claims system improvements for senior management.