Shashank Garg

One final college gig in the IoT domain that would revolutionize IoT developer space!

• Set up distributed application server infrastructure using sockets with Netty and Java.
• Responsibilities included setting up fault tolerant network topology, reactive leader election of group of servers, voting algorithms for achieving consensus for request handling, like PAXOS.
• Configured Replicated MongoDB database using replica sets for data redundancy and partial tolerance.
• Developed a unique algorithm for leader election called the “Assertive Bully… Show more

• Set up distributed application server infrastructure using sockets with Netty and Java.
• Responsibilities included setting up fault tolerant network topology, reactive leader election of group of servers, voting algorithms for achieving consensus for request handling, like PAXOS.
• Configured Replicated MongoDB database using replica sets for data redundancy and partial tolerance.
• Developed a unique algorithm for leader election called the “Assertive Bully Algorithm”.
• Mimicked a DNS server for leader address resolution amongst groups of clusters using Replicating HashMap.
• Set up chat server (blackboard collaboration) and file transfer service (homework assignments upload) as built in features.
• Set up secure RESTful end-points for service consumption using Ruby and Sinatra cookies and sessions.
Show less

• Created Real time Web application which tracks the sentiment of a unique hashtag in real-time and plots the results in a real time graph using D3.js
• Responsibilities included configuring and setting up a 4-node Hadoop cluster, including the general Hadoop services like MapReduce, Hive, HDFS as well as specialized data consumption services like Apache Flume.
• The Twitter fire hose stream was set up and was consumed by the Apache Flume service and persisted in the… Show more

• Created Real time Web application which tracks the sentiment of a unique hashtag in real-time and plots the results in a real time graph using D3.js
• Responsibilities included configuring and setting up a 4-node Hadoop cluster, including the general Hadoop services like MapReduce, Hive, HDFS as well as specialized data consumption services like Apache Flume.
• The Twitter fire hose stream was set up and was consumed by the Apache Flume service and persisted in the HDFS.
• Combination of machine learning algorithms like Naïve Bayesian, MaxEnt and SVM Is applied in the consumed tweets to extract the sentiment as positive, negative or neutral.
• Other lexical analytical methods like Recursive Tensor Neural Networks were also applied for complex and compounded sentences to handle negation, etc.
• HBase and Hive are then used to query the results periodically and plot the graph using D3.js
• Other responsibilities also included maintaining the load on the Hadoop cluster such that it may continuously consume new data and plot the results, along with archiving redundant data to conserve the resources.
Show less

- Plan your event is Web based application to create your own customizable event like Birthday, RoadShow etc
- Technology Stack: REST, JSON, Dropwizard, Jongo, MongoDB, Free Marker Template, STOMP, ActiveMQ

An independent project for Computer Services center of my institution.
Use Cases: To let the students and employers use the application to contact, filter and create events to talk and recruit students.
I implemented the resume building module and developed the notification module to inform the stakeholders.

Technology Used: JSP Servlets, MYSQL for backend.

Instant alerts of deals for a customer walking outside a store using Bluetooth sensing Gimbal beacons installed in a store. The deals can be updated in real time using a webapp built for store managers.

Technology Stack: Gimbal beacons, Android, Node.js backend, MongoDB, AngularJS

Android app to scan a barcode(UPC) of a product and fetch the prices it is selling for on various e-commerce websites.
Technology Stack: Android and Java REST APIs. Used Android Studio

Simulated an mobile app UI framework by developing from scratch to enable thinking/building using software design principles. Implemented 5 design patterns viz. command, state, factory, template, and adapter pattern.

Technologies used: Processing and Design Patterns.

a) Researched on the need for creating automated machine learning systems that sift through the big data sets and discover patterns without deep human intervention.
b) Projected minimum 50% decrease in data analysis cost by using Intelligent Big Data systems.

https://www.dropbox.com/s/haxjs1e4ra0lce1/IntelligentBigDataAnalytics-Mavericks.ppt

Developed an application, which takes 20 signature samples for training the ANN, classifies and differentiate with the forged ones using Back-Propagation algorithm. The accuracy achieved was 94%.

Technology used: Matlab 5.0