Karan Sahni Email and Phone Number

As the Assistant Manager at _VOIS, I worked on multiple important projects:1) End-to-end machine learning pipelines (from development to production) using GCP Vertex AI for Ireland and Spain markets-Developed robust data preprocessing workflows, addressing data quality issues and ensuring that the input data met the model's requirements-Designed ML models using standard practice of scikit-learn and deployed them on Vertex AI for training and evaluation-Tools used: GCP Vertex AI, Python2) Customized Auto ML Product for Forecasting- Developed a customized product for forecasting using Darts package to serve as an auto ML model- Embedded missing value treatment, outlier treatment, multiple models for time series forecasting, model selection via cross validation- Algorithms used: Auto ARIMA, Exponential Smoothing, Regression models, Naïve Mean, Random Forest-Tools used: Python3) MLFC – Machine Learning and Forecasting for Strategic Decision Making- Leveraged Machine Learning and forecasting algorithms to generate Monthly Performance Reports with insights for multiple Vodafone entities - Executed end-to-end automation for Turkey Consumer Business Unit (CBU) forecasting projectAlgorithms used: ARIMA, XG-BOOST, RANDOM FOREST, KNN, SMA-Tools used: Python, Alteryx4) Anomaly Detection- Took over Balance Sheet Reconciliation (BSR) tool to highlight risky General Ledger accounts - Contributed to significant reduction in manual effort and greater efficiency Algorithms used: IQR, GESD-Tools used: R

- Underwent rigorous training on Data Modelling and warehousing, Data analysis and Visualisation as a Post Graduate Trainee (PGT) under Data Analytics and Insight Services (DAIS)- Applied these skills on a live project for Care Operations where historical outage minute data were leveraged to provide insights for improved network availabilityAlgorithms used: Linear Regression, Time Series Analysis

- Analyzed subscriber data for three months in the Maharashtra and Goa (MNG) circle- Calculated demand price elasticity of subscribers given the talktime and data usage - Recommended customised subscription plans for non-unlimited users by leveraging a time series model

- Spearheaded brand building, explored new channels for revenue generation and established connections with target audience- Led a team of 10 interns and managed social media platform for promotion- Selected as Leader (intern) for the year 2017-18

- Detailed analysis of Portfolios for clients- Thorough knowledge of stocks and bonds to provide financial guidance