Murli Singh Rajpurohit

MURLI SINGH RAJPUROHIT 17016, PSN-Whitefield, Bangalore, India • - •-GitHub : https://github.com/murlisingh LinkedIn : https://www.linkedin.com/in/msr5850/ SUMMARY Meticulous Data Analyst with 3 years of experience interpreting and analyzing data in order to drive successful business solutions. Winner of “Excellence in HR Analytics” award. Achievements include creating data classifier model to predict company employee attrition with 23% more accuracy than historical average. Highly skilled in Machine Learning, Data Visualization, and creative thinking. EDUCATION Acharya institute of Technology – Affiliated to Visvesvaraya Technological University 2017 Bachelor’s of Engineering – Electrical and Electronics 75% (First Class with Distinction) • Maintained First Class with Distinction throughout engineering. • Final Year Project on “MATLAB/Simulink based performance analysis of Micro-Turbine Generator (MTG) for Grid/Islanding mode” which used Artificial neural network controller (ANN) for voltage variation and turbine governance. • Relevant Coursework: Signals & Systems, Digital Signal Processing, Control Systems, Operational Research, MATLAB Programming PROFESSIONAL EXPERIENCE Lenovo India HR Analytics - Data Analyst • • • • • Bangalore, India Sept’2018 – Present Built employee attrition prediction model in python using Catboost classifier with accuracy of 94.3% improving employee retention by 8%. Automated Workforce Planning process for Asia Pacific region (15+ Countries) on Tableau, leading to a 80% reduction in manual reconciliation time. Created Return-on-investment (ROI) and Business Performance dashboard including Workforce Productivity & Labor E/R, resulting in easy decision making for business stake holders. Constructed HR Chatbot named “Genie” using RasaNLU in python to answer Workforce related questions, resulting in 50% more employee engagement. Designed and built sales prediction model on large data sets using Keras and LSTM achieving 88% accuracy that helped in planning budget and target for next year. Starcom Infotech Bangalore, India Visual data analytics company providing solution in the data quality, self-service business intelligence & customer 360 view Data Analyst • • • • • Dec’2017 – Sept’2018 Improved the existing reporting dashboards and the functionality of planning tools. Reduced data processing time by 65%. Assisted product team to incorporate Natural Language Querying (NLQ) & Time Series Forecasting Models like ARIMA, ETS, TBATS, HOLT Winter into Star BI. Boosting sales by 20%. Carried out POC and demo of Data Quality and Business Intelligence products for various clients that deal with Retail, Finance, Banking, Education, Telecom etc, resulting in acquisition of Pfizer, Bharti AXA, Sanlam wealth management generating $400k USD revenue per year. Built Cross-sell model in python using library “Orange” and apriori algorithm achieving high Lift Ratio for Bharti AXA resulting in 15% more revenue. Designed 3 onsite databases and maintained a group of 25 databases. Through automation, improved efficiency by 15%, freeing up 50 labor hrs/mo. RESEARCH EXPERIENCE SystemonSilicon Corporation (Remote) San Francisco, USA Product based startup providing solution in Climate-Smart & Precision Agriculture and Personalized Digital Health Analytics Research Intern – Data Science • • • • Aug’2017 – Dec’2017 Designed crop disease detection model using DCNN based on AlexNet Architecture achieving accuracy 95.1% Solver type: Stochastic Gradient Descent, Base learning rate: 0.005, Learning rate policy: Step (decreases by a factor of 10 every 30/3 epochs), Momentum: 0.9, Weight decay: 0.0005, Gamma: 0.1, Batch size: 100. Created database back-ends for three mobile apps - AgroTick, MenGo and RiteFood. Built automated ML model to detect Bacterial test strip and crop it from a given image using OpenCV and contour detection, resulting in 95% reduction of manual detection and cropping time. Built Chatbot using IBM Watson to handle diverse health related questions by specifying intent, entities and creating a dialog flow. PUBLICATION Packt Publishing Mumbai, India The leading provider of technology ebooks, video tutorials, books and articles based in United Kingdom (U.K) & India. Title - “Hands-on data visualization with Microsoft Power-BI”. • • Dec’2018 Authored this video course to help BI consultant, Data analyst and other analysts to analyze and get insights from their data to support decision making. Sold 2000+ copies and received a positive feedback from course takers. TECHNICAL SKILLS • • • • • • Data Management: MySQL, MS-SQL, Mongo-DB Reporting Tools: Tableau, Power BI Programming: SQL, C, Python, R, Matlab Statistical Tools: Python, R, MS Excel, SPSS Machine Learning: Tensorflow, Keras, OpenCV, Numpy, Pandas, Pyspark, Scikit-learn, Neural Networks, CNN, RNN, Linear Regression, Logistic regression, Random Forest, Decision trees, Ada Boost, SVM, KMeans Clustering, PCA Natural Language Processing: Regex, NLTK, Gensim, Spacy, Polyglot, RNN, LSTM, Attention model, DSSM, NLU, Latent semantic analysis, SENNA, BOW, Glove, Knowledge Base, DQN, DRRN, Visionlanguage multimodal intelligence, Stacked attention network AWARDS & ACHIEVEMENTS • • Received an award in the category of “Excellence in HR Analytics” from The Society for Human Resource Management (SHRM) – a professional HR membership association headquartered in Virginia (USA). Qualified for Combined Annual Training Camp in National Cadet Corps at regional level. PROJECTS Image Classification based on Clothes length Dec’2017 – Feb’2018 • Analyzed an image data set and classified it into 3 categories: Full sleeves, Half sleeves, Sleeveless. • Used Keras with tensorflow as Backend. • Filter size (3 x 3) with 20 Epochs and 0.25 dropout ratio. • Improved accuracy by tuning Hyperparameters like Dropout ratio to get accuracy of 93.5%. Loansmart Oct’2017 – Feb’2018 • Used Linear Regression algorithm to predict the interest rate given by a certain bank. • Tested the violation of Linear regression assumptions (Homoscedasticity, Cook’s distance, Error independent) and fixed it. • Got Adjusted R-square 0.84 and RMSE on test 2.13 Medical appointment no-shows Oct’2017 – Dec’2017 • Calculated correlation and used various ML classifier algorithms to find whether a person will show up on appointment date or not. • Implemented principal component analysis to extract most influential predictors. • Designed logistic regression, decision tree, random forest, SVM, Ada Boost models. • Benchmarked all algorithms, Ada Boost gave most accurate result with 0.81 area under ROC curve. CERTIFICATIONS • • • • Microsoft Research certified in Natural Language Processing. Geo University Professional certification in Digital Image Processing using OpenCV. How to build a Chatbot without programming on Coursera by IBM. HR Analytics using Employee data certification on DataCamp. EXTRA CURRICULAR ACTIVITIES • • • • Secured 4th Place in Inter-College Weightlifting championship lifting 140Kg (Snatch - 60kg, Clean&Jerk 80kg) under 69Kg weight category. Participant of National Cadet Corps affiliated by Ministry of Defence in India. Fix potholes on the road as it is a major cause of fatal accidents in India Visited Karunashraya (Project by Indian Cancer Society) often to assist in the daily functioning of the hospice.