Nitesh Swaroop

Experience

• 

  Indian Institute of Technology, Delhi

  Apr 2015 - May 2016

  Master Thesis Project

  Title: Reducing Delhi Transport Corporation Buses Breakdown by Employing Better Maintenance ScheduleDescription: The objective function is made of two components. The first is catalyzing a weighted total maintenance hours (maintenance utility) for the buses that were pulled for maintenance during idle time, and the second is minimizing the weighted total number of maintenance hours for the buses that are pulled out of their scheduled service for inspection. Mixed-integer linear programming solver of MATLAB has been used to solve the defined problem. The solution to this problem gives the maintenance schedule for each bus that is due for inspection as well as the minimum number of maintenance lines that should be allocated for each type of inspection over the scheduled period. Proportional Hazard Model (PHM) is used to evaluate the effect of covariates on critical component failure and predict the failure mileage of critical components. Determined the optimum time of Preventive Replacement by using Age Replacement Model & Weibull++. Finally an app was developed on Matlab which is used to generate maintenance schedule for all the buses in a depot. Show less

• 

  Genpact

  Aug 2016 - Mar 2022

  PythonTurbine Part Classification and Defect Detection (Mar’19-Feb’22)● Worked on end-to-end project from getting the requirements from the client and getting it to the deployment.● Worked on Image Classification wherein the more than 100k images of various parts of the turbine have to be classified. Developed a classification model to classify 132 different turbine parts by using Pytorch, Tensorflow and Mxnet.● Performed Image Segmentation using Resnet-50, U-net, ENet, FCN, MultiScale methods-DeepLabv3 and PSPNet to segment 64 different damages on the turbine like crack, burn, oxidation etc. ● Performed training and testing of the model on AWS Sagemaker.● Encapsulated the model in docker and deployed it on AWS EC2 gpu-instance. Aero turbine outage, parts, and resource forecasting (Aug’17-Dec’21)● Worked on time series data to forecast upcoming events based on the historical data. The methods used are LSTM, Random Forest ,Regression, Holt – Winters, Exponential Smoothing, Gradient Boosting and SARIMA.● Predicted the turbine part demand for the vendors by using the failure rate, reliability and probability of failure. Distribution fitting is done on historical data to determine the failure parameters. ● Developed front-end by integrating python with Tableau for better visualization. Text Analytics (Aug’16-Jan’22)● Implemented Natural Language Processing to classify, categorize, organize and cluster the records. The developed model uses Support Vector Machine to classify the scale of turbine damage by analyzing the concern entered by engineers. ● Worked on Text Extraction Projects wherein the main objective is to build the code to answer multiple questions thus summarizing the contract.The procedure involves use of OCR and PyPDF2 to get the data and then export the output to Excel.Chatbot (Dec’20-Mar’21)● Worked on chatbot development using Amazon Lex and exploring SpaCy, BERT, GPT-1,GPT-2 Show less PythonAero turbine outage, parts, and resource forecasting (Aug’17-Dec’21)● Worked on time series data to forecast upcoming events based on the historical data. The methods used are LSTM, Random Forest ,Regression, Holt – Winters, Exponential Smoothing, Gradient Boosting and SARIMA.● Predicted the turbine part demand for the vendors by using the failure rate, reliability and probability of failure. Distribution fitting is done on historical data to determine the failure parameters. ● Developed front-end by integrating python with Tableau for better visualization.Tableau (Aug’16-Mar’22)● Comprehended the requirements and functionality of the application from ETL (Extract, Transform & Load) Developers, Project Managers and Members of the QA Teams● Created Tableau data visualization using Cross Tabs, Heat Maps, Box and Whisker Charts, Scatter Plots, Geographic Map, Pie Charts, Bar Charts & Density Chart● Developed donut charts and implementing complex features in charts like creating bar charts in tool-tip● Synchronized the SQL queries for maximum productivity and performance● Produced context filters & data source filters while managing huge volume of data● Utilized Tableau Server to publish and share the reports with the business usersR (Aug’16-Dec’16)● Integrated Tableau with R to leverage the statistical libraries of R● Built forecasting model to predict the upcoming outages based on past data Show less

  • Assistant Manager

    May 2019 - Mar 2022

  • Business Analyst

    Aug 2016 - May 2019
• 

  ExxonMobil

  Mar 2022 - Sept 2022

  Data Specialist

  Working on safety data from sensor to predict upcoming incidents(accidents) by implementing text analytics and predictive analytics. Performing model training and deployment on Azure. Also, working on multiple automation projects along withTableau dashboard development and enhancements.Python● Working on to classify free hand text based which describes the incident into severity level of categories using NLP● Used Azure to automate task which converts handwritten forms to excel or text.● Working to automate manual task which involves preparation of excel report from multiple data sources. Show less

• 

  Sun Life

  Dec 2022 - Sept 2023

  Machine Learning Engineer

  As a Machine Learning Engineer, I specialized in automating model training and deployment using Amazon SageMaker. My responsibilities included designing and optimizing machine learning models, preparing data, deploying models, and ensuring scalability.

• 

  GE Vernova

  Sept 2023 - now

  Lead Analytics Operations Specialist