Minesh Kumar Email & Phone Number

Chicago, Il, Us

Riverwoods, Il, Us

Cincinnati, Ohio, Us

Teaching Assistant for Data Mining in R

Lincolnshire, Il, Us

Directed data science efforts for demand forecasting for Hospital Patient Monitoring division of a Healthcare client- Generated a 4-month ahead forecast with 83% accuracy, using LSTM, ARIMA and Holt winters- Created data cleaning and forecasting pipelines on Azure MLFLow; designed and executed Outlier Detection method using STL decomposition- Implemented Data Triaging to prioritize training on high-impact data for maximum accuracy improvementContributed to the development of the in-house product for easier ML experimentation- Developed the backend for SynapseML implementation of LGBM, authored data processing modules foroutlier correction, time series disaggregation- Conducted rigorous A/B testing, leading to more accurate and robust forecasting methods; this approach informed feature refinement and model tuning, resulting in a 10% reduction in forecast errorWorked on Demand Forecasting at scale for a consumer appliances company, improving the company’s on-time delivery indicator from 0.3 to 0.5- Used LGBM for forecasting; achieved accuracy of 59%, enabling low-touch demand planning for high-selling products that contributed 60% to the annual revenue- Developed data cleaning, feature generation and prediction scripts on Azure Databricks; automated the monthly forecast generation process using Azure MLFlow- Performed extensive EDA on transactions, promotions, and lineage data; incorporated business checks to correct data for the disruption caused by COVID-19- Conducted Forecast Output Analysis on PowerBI to identify areas of improvement in modelling

Marlborough, Massachusetts, Us

Built a claim likelihood prediction model using XGBoost for an insurance client; achieved a recall rate of 85% and precision rate of 72% in identifying potential claims- Optimized XGBoost performance through hyperparameter tuning using caret package and performed cross- validation to obtain the best generalized model- Set up an Apache Zeppelin server on a GCP VM for data preprocessing, feature engineering and data retrieval- Deployed the model as a REST API using plumber library, enabling easy integration and usage by the client Performed descriptive analytics to analyze the impact of underwriting changes on an insurance client’s business- Utilized Apache Zeppelin, plotly, and Google Data Studio to generate visual reports- Troubleshot and resolved technical issues, including solving a dependency that prevented interactivevisualizations from rendering in Apache Zeppelin

Master Of Science - Ms, Business Analytics & Data Science

Bachelor Of Technology (B.Tech.), Civil Engineering