Software Engineer - Data Sciences
- Return item quality predictor model - The target feature recognizes an abused return item and helps Myntra decide whether to do 'Inward Quality Check (IQC)' or not - hence, saving on cost and manpower and reducing average return refund time by half. Currently in pilot across 10 delivery centers in India.
- Consumption NPS drivers analysis - The consumption NPS is a metric to define customer satisfaction with product usage. I developed a model using random forest and K-Means clustering to derive drivers for this metric and profile the characteristics which lead to a satisfied customer.
- Churn prediction model - The model gives the probability of churn in a quarter for Myntra customers. Features included delivery experience, cancellations, returns experience and CC interactions
- Pained customer redressal framework - Quantified the impact on customer transaction frequency due to various service/experience short-falls. On the basis of which designed a framework for redressing pained customers and improve customer retention.