• Twitter User's Country Prediction

    Worked with 2 others to predict country from which user belongs, based on his tweets, tried and tested various machine learning algorithms during the process.

  • Quora Question pair duplicate detection

    Project to detect if 2 questions sematically means the same thing. Based on Data set released by Quora
    Tried 4-5 approaches from basic to deep learning BiDirectional LSTM.

  • Adverse drug interaction effect

    Building Automatic system to Detect if the symptoms described by the patients are due to the adverse reaction of the 2 drugs which patient is already taking.

  • Word2Vec Implementation

    Single layer neural network implementation based on Cross Entropy and NCE for finding the best vector representation of the words.

  • Dependency parser - deep learning

    Implementation of the Dependency parser as per the paper "A fast and accurate Dependency Parser using Neural Network" and different experiments with the approach in order to explore different setups with their effect on the accuracy.

  • Relation Extractor - deep learning

    Performed different type of feature extraction tasks for the given training and test data in order to predict whether the given sentence contains "educated from" relation or not including feature extraction from the dependency parse tree of the sentence.

  • Merge system call implementation

    Implementation of merge system call which merge sort 2 sorted files and generated the output file.

  • Stackable file system

    Individually worked on developing a system to trace all user's operations on files, added via stackable filesystem, traces stored in non-readable repayable file, ioctl support to customize the system call listing to trace.

  • Custom System call Support

    Worked with 2 others on developing framework in linux kernel to support custom implementation of system calls, a process can notify kernel to use default or custom system call vector

  • Yelp Data Analysis

    Individual project on Developing Page Rank algorithm to find out most popular businesses. Used numpy, pandas, Ipython, Scipy, Scikit-learn, matplotlib

  • This website

    Designed and implemented this website.

  • Immigration visualization

    Visualization of how people immigrate across different countries in the world. Include interactive visualizations of the data from past 10 decades.

    Final Product

  • D3 Interactive plots

    My first try to implement a interactive visualization in D3. Provide a dynamic bar chart and pie chart visualization of basket ball player data.

  • PCA, MDS visualization using D3

    IMDB movie data analysis including dimension reduction from 10 to 2 using PCA and MDS. plotting dimensionally reduced data and analysing both the approaches.

  • Built 3D Printed CNC Mill

    As part of this project we built a CNC Mill. Used Laser cut and 3D printed parts. wired up the electronics and used grbl for software communication with the machine.

    Final Product

  • Web Application Firewall

    A reverse proxy based application which sits in front of web server and protect it from malicious requests by finding defined patterns in the requests as well as learning and blocking anomalies automatically.