Animesh Pandey
Interested in NLP, Search-Engines and Machine Learning
Software Development Engineer at Audible Inc. (an Amazon company)
LATEST PROJECTS
Hub-Academia (Mini Forum)
Aug 2016 - Oct 2016
DESCRIPTION: Mini framework designed for understanding the MEAN Stack. It uses Google Books API for searching books on Google. I also implements oAuth2.0 for logging in using Google.
LANGUAGE: NodeJs, Angular 1.5, Express, MongoDB
FRAMEWORKS/LIBRARIES: Google Books API, Passport
VADER Sentiment Analysis
March 2016 - April 2016
DESCRIPTION: Valence Aware Dictionary and Sentiment Reasoner or VADER is a rule based sentiment analysis technique which is included in Python NLTK. This project is a Java port of that library.
LANGUAGE: Java
FRAMEWORKS/LIBRARIES: Lucene (Text Processing)
ADVERTISING FRAMEWORK
Final Year Project
July 2012 - May 2013
DESCRIPTION: A framework, in which a user can automatically embed advertisements in a video and save it, was designed. Initial prototype was designed in Visual C++ using OpenCV to test the algorithm that was designed. The object detection techniques tested are Haar Classification and Bag-of-Words Model. We have tested 4 objects in a 17 second customized video.
A patent has also been filed on this framework. Patent application number is GB1305811.0 and filing date 28.03.13.
LANGUAGE: Visual C++ (Visual Studio 2010)
FRAMEWORKS/LIBRARIES: OpenCV 2.4
Post Final Year
June 2013 - Oct 2013
DESCRIPTION: Currently, we are re-programmimg it in Python to make its code more readble and platform independent.
LANGUAGE: Python 2.7.3 (PyScripter)
FRAMEWORKS/LIBRARIES: OpenCV 2.4, Python Image Library (PIL)
QUERY RECOMMENDATION AND TOPIC MODELLING
Natural Language Processing Semester Project
Jan 2013 - May 2013
DESCRIPTION : Correct recommendations to the users is an important part of every internet based business. When a user posts a query on a forum, sometimes he is asking question that has already been asked or he is asking a very popular question. If the system is able to display the top most popular question or the questions that have similar context to that of the qeury intended to be asked, the user might not waste time in posting that and the website people will save space form redundant questions.
Next, part was the automatic tagging of the questions asked as well as stored. Topic Modelling was done by using Latent Dirichlet Allocation by training the classifiers using corpus of 40,000 questions that was collected from StackOverflow.com using their API. Every question was tagged with 5 tags.
LANGUAGE(S) : Python 2.7.3 (PyScripter), CGI, Ajax
LIBRARIES : sklearn, numpy, scipy, gensim, nltk
MOVIE RECOMMENDATION SYSTEM
Information Retirieval and Data Mining Semester Project
Jan 2012 - May 2012
DESCRIPTION: Websites like Youtube, Netflix recommend videos/movies on the basis of the users profile, likes, history etc. We tried to emulate the algorithms that are used in recommendation systems. We had used the Movie Lens dataset that had data of around 1000 users and 100000 ratings of aroung 500 movies. We had used Nearest Neighbours method to recommend similar movies to the user. The similarity of the users was computed by the cosine similarity between the rating vectors. We were also able to segregate users on the basis of the genres of movies they prefer.
LANGUAGE: PHP/MySql, Ajax
FRAMEWORKS/LIBRARIES: None
ONLINE C/C++/Java IDE
Web Application Engineering Semester Project
July 2011 - Dec 2011
DESCRIPTION: Developed an online IDE where a user could write, load and save programs on the web browser itself. We used piplining to exectute the programs and display the output on the browser, We used GCC and Java 1.6 for the compilers to be used in the project.
LANGUAGE: PHP, Javascript, Ajax
FRAMEWORKS/LIBRARIES: None
OPTICAL CHARACTER READER
Algorithms Design Semester Project
July 2010 - Dec 2010
DESCRIPTION: Developed an Optical Character Reader in Matlab. This used data to train using Artificial Neural Networks. Thinning algorithm and similar image processing techniques were applied to the project. Image was input and result was text in a Notepad file.
LANGUAGE: Matlab
FRAMEWORKS/LIBRARIES: Neural Networks Toolbox