Engineer Data

ALEKYA KUMAR

412-***-**** LinkedIn : Alekya Kumar ********@*******.*** My Github

EXPERIENCE

GRADUATE RESEARCH ASSISTANT UNIVERSITY AT BUFFALO JUNE 2018 – DEC 2018 A research project on accidents that happen due to Vehicles on the National Highways and how they can be minimized by bringing forth changes to Urban Planning

· Collected Data from www.NHTSA.gov, integrated tables and, analyzed factors leading to death and reduced the dimensions using Feature Selection methods such as Multiple Correspondence Analysis and Stochastic Diffusion Search

· Developed Machine Learning Algorithms such as Poisson Regression to predict the Number of Deaths; Logistic Regression and, Decision Trees, to classify the data by the number of deaths and obtained an accuracy of 81% in terms of AUC-ROC characteristics SYSTEM ENGINEER TATA CONSULTANCY SERVICES DEC 2013 – JUL 2017 BAGGAGE SORTATION MESSAGE SYSTEM

An automated system that helps in efficient sorting and delivery of bags in order to reduce the wait-time of passengers

· Liaised with the client to implement an automated baggage sort system to deliver bags to respective terminals

· Involved in end-to-end data analysis and deployment of BSM thereby reducing time for baggage delivery by 70% PROJECTS

PREDICTIVE ANALYTICS EARTHQUAKE DAMAGE PREDICTION-HACKEREARTH DATA CHALLENGE PYTHON A Predictive Model to assess the extent of damage that was done to a building post-earthquake

· Performed EDA on the data; Conducted Chi-Square Test to understand their relationship with target variable

· Performed Feature Engineering and fitted Decision Tree and Random Forest models and produced an accuracy of 84.1%, the third highest in the competition

MACHINE LEARNING HANDWRITTEN DIGITS AND FACE RECOGNITION PYTHON A Supervised learning model implemented using Neural Networks to recognize handwritten digits and Celebrity Faces

· Implemented Forward Pass and Back Propagation algorithms for different combinations of hyper parameters and produced maximum accuracy of 90.2%

· Designed a Deep Learning model - Convoluted Neural Networks(CNN) using Tensorflow for recognizing celebrity faces; tuned the parameters and produced an accuracy of 97%

RECOMMENDER SYSTEM ONLINE DATING WEBSITE R, SPARK, AWS A system that provides recommendation to users for a Dating website

· Conducted initial analysis on the dataset, containing 17 million records/ratings given by users, using Spark and AWS and understood the ratings provided by each user to another

· Devised an algorithm that uses ratings and similarities between users to recommend user profiles. Produced a very low error of 2.6579 using Pearson Correlation methodology

EDUCATION

MASTER OF SCIENCE – DATA SCIENCE DEC 2018 UNIVERSITY AT BUFFALO

· GPA : 3.5 / 4

BACHELOR OF ENGINEERING JUNE 2013 ANNA UNIVERSITY

· GPA : 7.83 / 10

RELEVANT SKILLS

· SOFTWARE : Python, R, Tableau, SQL, MATLAB, MS Office, Unix and Shell Scripting, JAVA

· TECHNIQUES : Linear/Logistic Regression, Decision Trees, Time Series Analysis, Recommender Systems, Neural Networks ADDITIONAL

· Finalist in the challenge conducted by Data Science Education company, Data Incubator

· Certified in Deep Learning(Specialization) on Coursera – Deep Neural Networks & Hyper Parameter Tuning

Contact this candidate