Simran Chaudhari

Data Scientist

Capgemini•  October 2023 - Present•  Pune

Predicted mining asset health by forecasting machine behavior and collaborated with clients to pre-inform them about upcoming failures minimizing shutdowns and production losses. Developed a machine learning model using the XGBoost algorithm to predict the remaining useful life of the mantle liner in a gyratory crusher, allowing operators to proactively schedule maintenance activities, order replacement parts in advance, and optimize crusher performance. Achieved a maintenance cost reduction of $1.5M and decreased the number of maintenance shutdowns from 4 to 2 per year, with the model attaining an accuracy of 94.%. Created a machine learning model using Prophet to forecast the health of dust collector bags, providing a 21-day forecast. This model enabled regular monitoring of the dust collector’s health and allowed the extension of maintenance windows when bags still had remaining life based on model predictions.Achieved a model accuracy of 91%, reducing maintenance shutdowns from 4 to 2 per year. Developed a sophisticated text summarizer application leveraging Hugging Face Transformers to condense large volumes of text into concise summaries.Implemented state-of-the-art transformer models to achieve high accuracy in text summarization.Designed an intuitive user interface allowing users to input and summarize text effortlessly.

Senior Analyst

Capgemini•  August 2022 - October 2023•  Pune

Developed a predictive model for analyzing silica content in bauxite ore by implementing clustering algorithms (Agglomerative Hierarchical, K-means).Utilized Partial Least Squares Regression (PLSR) to accurately estimate silica content, improving spectral analysis and ore quality assessment. Implemented Quadratic Regression model to detect impeller wear out failures by calculating Remaining useful life (RUL) and forecasting Risk of Failure three weeks in advance. Successfully extended impeller component lifespan by over one year compared to traditional maintenance practices. Achieved a model accuracy of 90 % for impeller condition forecast leading to enhanced operational efficiency and cost savings. Developed a customized web application using Django framework for conducting preliminary data analysis and data visualization efficiently. Users can easily upload datasets in various formats (CSV, Excel) through a user-friendly interface.This tool enable users to perform EDA, including summary statistics, visualizations (histograms, scatter plots, etc.), and correlation analysis. This app help users gain insights into their data distributions and relationships. Created a Django Application that assesses the quality of Python and R code and generates comprehensive reports detailing code quality and providing suggestions for code improvement to enhance efficiency and maintainability using Pylint and Rlint libraries.