Information Technology Computer Science

* * * * *

GURU PRASAD NATARAJAN

www.gurunatarajan.com

********@****.***.*** 312-***-**** www.linkedin.com/in/gurunatarajan github.com/guruvasuraj PROFESSIONAL SUMMARY

• Computer Science graduate with 5 years of Software Development and Machine Learning experience.

• Experience uncovering hidden patterns from large sets of structured, semi-structured and unstructured data into actionable insights.

• Resourceful and flexible with the ability to adapt to changing situations and priorities; highly organized and effective in problem resolution and multitasking.

• Energetic and adaptable, a team player who is always willing to go the extra mile trying to solve the previously impossible. EDUCATION

ILLINOIS INSTITUTE OF TECHNOLOGY Chicago, IL

Master of Computer Science, with Specialization in Data Analytics Expected May 2017 ANNA UNIVERSITY Coimbatore, India

Bachelor of Technology, Information Technology May 2011 EXPERTISE

• Languages: Python, Java, Go, SQL, C, and C++.

• Web Application Frameworks: J2EE, Struts, Spring MVC, Hibernate, Flask.

• Python scientific stack: NumPy, SciPy, scikit-learn, pandas, Matplotlib, IPython.

• Deep Learning Frameworks: Caffe, Keras, Theano, TensorFlow.

• Statistical Analysis Tools: R, SPSS, Excel (Regression, Correlation, Data Visualization).

• Hadoop Ecosystem: Apache HDFS, Apache Pig, Apache Hive, Apache Sqoop, Apache Spark and Map Reduce.

• Version Control: GIT, Bitbucket, IBM Rational Clearcase, SVN.

• Certificates: Oracle Certified Java Programmer (OCJP) 1.6 and Microsoft Certification in HTML5, CSS3, and JavaScript. EXPERIENCE SUMMARY

Machine Learning Engineer at Mindboard Inc., – Chantilly, Virginia Oct 2016 – Present

• Developed deep learning models using Caffe, CUDA, and cuDNN for vRate - a chrome web extension which blocks offensive contents in websites (images, texts and videos) not suitable for younger audience.

• Wrote python scripts and automated the processes to ingest and cleanse data from multiple data sources.

• Preprocessed images using OpenCV which included resizing, rotation and random crops to generate rich training dataset.

• Performed skin detection analysis using RGB ratio model and classified the images using convolutional neural networks.

• Utilized AWS EC2 instances to host APIs and NVIDIA GPU rest engine to perform load tests on the models developed.

• Web scrapped text data using BeautifulSoup and developed python scripts to clean and prepare data for the model.

• Used Latent Dirichlet Allocation for topic modelling.

• Developed character-level convolutional neural network to classify textual from the websites.

• Working on an analytics tool to generate reports on a day-to-day basis to be sent to authorized personnel. Systems Engineer for Citibank – North America at Tata Consultancy Services – Chennai, India Nov 2011 – Dec 2014

• Analyzed BRD provided by the client and provided feasibility status for Citi Retail Services.

• Involved in the creation of Design Document, Project Estimation, and Delivery Plan.

• Designed and developed reusable components using Struts MVC and Spring MVC.

• Designed the user interface of the application using HTML5, CSS3, Bootstrap, JSON, JavaScript, JQuery and AJAX.

• Implemented client-side application to invoke SOAP web services.

• Developed SQL queries to store and retrieve data from database & used PL/SQL.

• Identified and implemented new tools for increasing team productivity such as back-end simulator and dashboard utilities.

• Used Ant scripts to build and deploy the applications in JBoss application Server.

• Closely collaborated with testing teams in identifying, analyzing and fixing issues to minimize bugs.

• Received various client accolades for delivering code with 0% post-production defects. Guru Prasad Natarajan ********@****.***.*** 312-***-**** 2 P a g e

PROJECTS

Hybrid Deep Neural Network for Text Classification Python, NumPy, keras, word2vec.

• Developed a Recurrent Convolutional Neural Network (RCNN) to classify texts without human-designed features.

• Used Recurrent Neural Network to capture contextual information and the max-pool layer of the convolutional neural network to determine the discriminative features to represent the text.

• Compared the new model with traditional text classifiers and neural networks.

• RCNN outperformed traditional models and other neural networks approach and attained 96% accuracy. Twitter Analysis on Marijuana Python, NumPy, Scikit-learn, pyplot, MongoDB.

• Used Twitter’s REST API to collect tweets related to marijuana using related keywords and dumped the data into MongoDB.

• Preprocessed the tweets and employed Logistic Regression to classify the sentiments of the tweets.

• Inferred the demographics of the users by named lists approach and K-Nearest Neighbors classifier.

• Predicted the sentiments of the tweets with more than 90% accuracy. Time Series Analysis Python, Pandas, Matplotlib

• Developed Recurrent Neural Network models to predict the number of outbound passengers from a certain airport.

• Created various time series regression models using time steps and window based techniques, and memory batch techniques.

• Analyzed the models based on the number of errors obtained. Sequence Generation using Recurrent Neural Network (RNN) Python, Theano

• Developed a generative model to predict the next word in the sequence given the preceding words.

• Worked on a word-level RNN based on LSTM and GRU to generate text sequences.

• Learned how RNN architectures provide solutions to the long-range dependencies problem. Univariate and Multivariate Regression Analysis Python, NumPy, Matplotlib

• Loaded single feature data sets and plotted them to understand the complexity of the problem and used a linear model to fit them.

• Used ordinary least squares loss function to train the model and computed train and test errors.

• Experimented with various polynomial models and selected a model.

• For multivariate data sets, linear regression was performed on higher-dimensional space and evaluated different mappings.

• Root mean square error was calculated for both training and testing sets, used Gaussian Kernel function to solve dual linear regression problem.

Multilayer Perceptron for Image Classification Python, NumPy, Pandas

• Developed a two-layer feed forward perceptron to classify the input image into six different categories.

• Used data from the UCI ML repository to perform the experiments.

• Derived the backpropagation equations and analyzed the performance by varying the number of hidden layers. Recommendation System Apache Spark, Pyspark

• Built a content-based recommendation system to recommend restaurants based on the ratings on Yelp.

• Employed collaborative filtering technique to fill missing entries and used alternating least squares algorithm to learn the latent factors.

• Trained multiple models and selected the best model based on Root Mean Squared Error. Affinity Analysis Python

• Developed market basket analysis for a supermarket to determine association rules.

• Employed a priori algorithm to generate new association rules at every step.

• Varied the frequent items threshold to obtain a new set of association rules. AWARDS AND ACHIEVEMENTS

• Star Employee award and On the Spot award for excellence at work at Tata Consultancy Services.

• Awarded Harbinger of the learning group for technical skills displayed during training at Tata Consultancy Services.

Contact this candidate