Python, Java, Algorithms, Data Mining, Machine Learning, Hadoop, Spark

Prakhar Dogra

571-***-**** https://github.com/prakhardogra921 ******@***.***

EDUCATION

Master of Science in Computer Science, May 2018

George Mason University, Fairfax, VA GPA: 3.8

(Relevant Coursework: Machine Learning, Computer Vision, Pattern Recognition, Mining Massive Datasets using MapReduce, Data Mining) Nanodegree: Deep Learning (Udacity) May 2018

Nanodegree: Machine Learning Engineer (Udacity) Feb 2018 Bachelor of Technology in Computer Engineering, July 2016 Delhi Technological University, Delhi, India GPA: 3.5 TECHNICAL SKILLS

Proficient: Python, Java, C/C++, Hadoop, SPARK, Tensorflow, OpenCV Familiar: HTML, CSS, JavaScript, R, SQL, Django, Amazon EC2, Google Cloud RELEVANT EXPERIENCE

System Design Intern, Cafex Communications, New York Jun - Aug 2017

• Developed an Intelligent document parsing system to extract relevant information from meeting minutes

• Deployed the system online using Django web framework Data Science Intern, Premier Logic, Noida, India Jan - July 2016

• Scraped news websites in order to find recent news of a particular domain.

• Developed an API for Virtuagym to find clients interested in fitness and gym using Twitter. PROJECTS

• Bank Marketing Classification.

• Implemented various models like Decision Tree, Random Forest, Adaboost, Naïve Bayes, Multi-Layer Perceptron and Support Vector Machine.

• Conducted comparative study on the above models when applied on different feature sets obtained via feature selection (Ch- Square Test), feature transformation (Principal Component Analysis) and feature elimination (Recursive Feature Elimination).

• Youtube Video Label Classification.

• Implemented a Long term Recurrent Convolution Network to classify YouTube videos into multiple classes

• Extracted frames from videos as input to a Convolution Neural Network for feature extraction.

• The features so extracted are passed to a LSTM for predicting labels.

• Finding similar images using Locality Sensitive Hashing.

• Performed shingling on images followed by Min Hashing and Random Hyperplanes

• Generated Signature Matrix, a compact representation of images using above techniques.

• Generate candidate pairs from signature matrix using locality sensitive hashing in Apache Spark

• Anomaly Detection on Credit Card Fraud.

• Implemented Random Forest, log likelihood and various other models for detecting frauds.

• The dataset used for this project was Credit Card Fraud Detection dataset from Kaggle

• Meeting Minutes Parser.

• Developed a document parsing system and deployed it online, using Django web framework, that can extract required information from meeting minutes with a feedback system.

• Performed text analysis on meeting minutes to extract meaningful actions discussed in meetings

• Deployed the results on a web portal built with Django framework

• Social Media Complaint Workflow Automation Tool Using Sentiment Intelligence.

• Developed a complaint classification and forwarding mechanism for bank posts obtained from Facebook web pages.

• Used Twitter and Facebook graph API for building a crawler

• Used bag of words and NLTK for extracting features from text obtained from websites through crawler

• Classified posts into complaints, inquiries and irrelevant posts, then an additional layer of classification was added to classify complaints into relevant departments.

• Twitter Analysis to Find New Gym Clients.

• Used Twitter API to fetch tweets

• Implementing classifiers such as SVM, Multinomial Naïve Bayes Classifier and its variations

• After classification, actual usernames, mentions on twitter for the classified users were extracted.

Contact this candidate