Sign in

Engineer Machine

Durham, NC
September 10, 2018

Contact this candidate



Highly motivated and creative individual driven by passion with 2 years of experience in IT industry. Enthusiastic about Data Science and Machine Learning. Actively seeking full time opportunities.

217-***-**** Durham, NC EDUCATION

The University of Illinois at Springfield

Master’s in Computer Science (Data Science), May 2017 Guru Nanak Dev University, Amritsar, India

Bachelor of Technology in Computer Science and Engineering, June 2015 TECHNICAL SKILLS

Machine Learning models: Linear Regression, Logistic Regression, SVM, Naïve Bayes, K-Means, Random Forest, KNN, Gradient Boosting, XGBoost

Deep Learning models: Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), TensorFlow, Keras Big Data Analytics: Apache Spark (PySpark, Scala), Hadoop, MapReduce, Hive Programming: Python, R, Java, SQL, Scala, SAS, C++, C Relational Databases: Oracle 11g, MySQL, MS SQL Server, Teradata NoSQL Databases: MongoDB, CouchDB, RIAK, HBase, Neo4j Business Intelligence tools: Tableau, Jupyter Notebook, RStudio, MS Excel WORK EXPERIENCE

Computer Task Group Raleigh-Durham, NC (July 2018 to Present) Tools: PySpark, Python (Pandas, NumPy, Sci-kit Learn, SciPy, Matplotlib, Seaborn), Flask Machine Learning Engineer Intern

Currently working with client Lenovo to predict computer crashes by analysing data and training machine learning models. Implementing feature engineering, feature selection, explanatory data analysis and cleaning data before evaluating machine learning model’s performance. Dallas, TX, (June 2017 to July 2018)

Tools: Python, Artificial Intelligence (TensorFlow, Keras, Pandas, NumPy, Sci-kit Learn, SciPy) Data Scientist (Contract)

Successfully implemented Hog Classifiers on OpenCV to collect features of different objects and training Support Vector Machines (SVM) on AWS EC2 to classify the objects (vehicles, pedestrians and bikes) on the NVIDIA Jetson TX2

Implemented transfer learning and feature extraction on VGG to classify traffic signs with a validation accuracy of 96.12% and test accuracy of 90%

Behavior cloning of car using Deep Neural Net architecture with a combination of 5 Convolutional layers and 4 fully connected layers to predict the steering angle of the car at any given point of time with a validation loss of 0.0183 Nimbus Software Jalandhar, India

Tools: PHP, JavaScript and jQuery, MySQL (backend), PL/SQL, T-SQL (May 2015 to December 2015) Software Engineer

Worked on Nimbus Campus ERP educational institutes and colleges

Developed a web application for an NGO and centralized the data access logic using stored procedures and triggers PROJECTS AND RESEARCH

Kayak Rental Project (Machine Learning and Predictive Analytics) Tools: Python (Pandas, NumPy, Sci-kit Learn, Matplotlib)

Developed and trained a machine learning model for Kayak Rentals using Gradient Boost Classifier, reducing the overall root mean square error (RMSE) to 3.29, thus suggesting an increase in the fleet size by 7% to meet customer demand Movie Lens Recommendation System (Machine Learning and Optimization) Tools: Apache Spark, Scala (ALS, MatrixFactorizationModel, Rating)

Designed Item-Item based Collaborative filtering for Movie Lens Dataset from Kaggle using Utility Matrix and Pearson Correlation with the root mean square error (RMSE) 0.94

Suggested optimization using Latent Factor models like Singular Value Decompositions versus Gradient descent and including bias to bring down the root mean square error from 0.94 to 0.89 New Approach to Sentiment Analysis (Deep Learning) Tools: Python (Keras)

Designed a Recurrent Neural Networks (RNN) on data from customer complaints and feedbacks for given grocery store products, to classify customer likelihood with a validation accuracy of 90.28% using Binary cross-entropy and LSTM cells for long-term dependencies

Who is my Voter? (Machine learning and Predictive Analytics) Tools: R (Carat, irr, e1071, C5.0, plyr, gmodels)

Developed a machine learning model to classify American voters into republicans or democrats using Ensemble Learning with accuracy of 85.02% using Naïve Bayes and an accuracy of 92.10% using C5.0 KAGGLE COMPETITIONS

Tools: Python (Pandas, NumPy, Sci-kit Learn, SciPy)

Successfully classified Porto Seguro’s Safe Driver to predict the probability that whether the driver will initiate auto insurance for the next year or not using Machine Learning Algorithms like RandomForest and Xgboost Classifier

Contact this candidate