Pramod P. Setlur

Idea:
This project was in a hackathon in San Francisco. We tried to create a win-win eco system for customers and retailers using Beacons.

We have Beacons placed in bars or restaurants. The iPhone App can capture the deals or information sent out the Beacons (received via, BLE - Bluetooth Low Energy). The applications send the data to the cloud based REST service. This piece also processes the data and stores it in a DynamoDB

Going back home, we can ask Alexa to read out the… Show more

Idea:
This project was in a hackathon in San Francisco. We tried to create a win-win eco system for customers and retailers using Beacons.

We have Beacons placed in bars or restaurants. The iPhone App can capture the deals or information sent out the Beacons (received via, BLE - Bluetooth Low Energy). The applications send the data to the cloud based REST service. This piece also processes the data and stores it in a DynamoDB

Going back home, we can ask Alexa to read out the deals we collected and use it for something during the week or weekend.

Technical Details (A better view of the design is in the Github link) :-

Android app emulating data transfer with Beacons
iOS app to collect the data and push it to the cloud based Rest service
REST service, consumes and processed data, and pushes to DynamoDB
An Alexa skill, that pulls data from DynamoDB and returns it to the user. Show less

We a team of 6, developed a hyper-targeted ad system in the 24h hackathon. The project can be categorized as an extension to 'Community Detection'.

Technologies used: AWS, node, elastic, java (springboot)

More info:
We ensured that the ads displayed on the ticketmaster website was more personalized.

Examples
A person X, with a few tickets to a basketball game would have "Season Tickets Ads" displayed.
A person Y, with a season pass to a basketball game would be… Show more

We a team of 6, developed a hyper-targeted ad system in the 24h hackathon. The project can be categorized as an extension to 'Community Detection'.

Technologies used: AWS, node, elastic, java (springboot)

More info:
We ensured that the ads displayed on the ticketmaster website was more personalized.

Examples
A person X, with a few tickets to a basketball game would have "Season Tickets Ads" displayed.
A person Y, with a season pass to a basketball game would be displayed "VIP tickets".
A person Z, with no interest in sports, would be displayed an ad of Jimmy Buffet or an ad of a theater show depending on the person's score against various categories.

PS: The project is hosted on ticketmaster's internal git repo. It contains intellectual property and cannot be exposed on github. Show less

Project Description: It is an inventory management system that stores all the information about our on-prem VMs. hypervisors, products, and cloud. A Spring and React.js application powered by postgres and Elastic Search (search layer on top of postgres) in the backend.

My role: Automation / build engineer. I setup the build and deployment pipeline for the project with Continuous Integration and Continuous Delivery.

The build pipeline ensures automation from a git commit to a push… Show more

Project Description: It is an inventory management system that stores all the information about our on-prem VMs. hypervisors, products, and cloud. A Spring and React.js application powered by postgres and Elastic Search (search layer on top of postgres) in the backend.

My role: Automation / build engineer. I setup the build and deployment pipeline for the project with Continuous Integration and Continuous Delivery.

The build pipeline ensures automation from a git commit to a push of the docker image to ECR.

The deployment pipeline is setup using Terraform. The terraform template deploys the app on AWS. The infrastructure (EC2 instances, security groups, IAM roles, ELB - load balancer, etc) is built from scratch and pulls down the docker image and starts the app.

Accomplishment: A complete CI with jenkins. 1 click deployment. Zero downtime deployment strategies and implementations Show less

HackSC 2015, Winners

Video of the project: https://www.youtube.com/watch?v=dZLE14s5pGE
Github: https://github.com/ShantanuBal/AlexaShowFinder
Architecture: https://goo.gl/photos/95wF9iKACfjBYGKj8
Technologies used: AWS Lambda, AWS DynamoDb, Microsoft Azure


Developed ShowFinder application (alexa skill) on Echo device. Echo is a voice command device from Amazon.com. It connects to Alexa, a cloud-based voice service, to provide information, answer questions, play music,… Show more

HackSC 2015, Winners

Video of the project: https://www.youtube.com/watch?v=dZLE14s5pGE
Github: https://github.com/ShantanuBal/AlexaShowFinder
Architecture: https://goo.gl/photos/95wF9iKACfjBYGKj8
Technologies used: AWS Lambda, AWS DynamoDb, Microsoft Azure


Developed ShowFinder application (alexa skill) on Echo device. Echo is a voice command device from Amazon.com. It connects to Alexa, a cloud-based voice service, to provide information, answer questions, play music, read the news, check weather and more—instantly.

The ShowFinder application/skill on Amazon Echo looks for events/shows happening in the city, gives more details on the interested event and can hail an Uber if the user is interested in attending a particular event.
ShowFinder uses :
i) TicketMaster API's to search for the events/shows in a particular city on a particular date.
ii) Uber API's to book an Uber to the interested event. Show less

In this project I implemented a user-based collaborative filtering recommender system for predicting the ratings of an item using the sample data. This prediction is using k nearest neighbors and Pearson correlation. Finally using the similarity of the k nearest neighbors, I predicted the ratings of the new item for the given user.

Given a transaction file I implemented 3 algorithms (PCY, Minhash, Toivonen) to find the frequent itemsets. Each algorithm is an improvement in space and time as the previous one in the order above. The implementation and the README is well documented on the github link.

A Map Reduce project to solve a few daily problems. Matrix multiplication was solved in single phase and multiphase. A couple more were to find frequent itemsets (an application for this is in Recommendation systems) and finding term frequency and the document frequency.

I am working as a Software Developer in a large team of 40 - 50. We are to port the TIcketmaster website on AWS.
The project is built in a Docker container using Jenkins. Rundeck is then used for deployment.

I am developing a build-pipeline for continuous delivery of web-services. A git push of the code invokes a build on Jenkins. On a successful build a Rundeck Job is triggered which deploys into a cluster for smoke testing. This can be followed serially or parallelly with a regression test (using Gatling) against a different cluster.
A major subproject: Developing an end point for importing job templates to Rundeck. A POST request (body containing the job template) to this end point will… Show more

I am developing a build-pipeline for continuous delivery of web-services. A git push of the code invokes a build on Jenkins. On a successful build a Rundeck Job is triggered which deploys into a cluster for smoke testing. This can be followed serially or parallelly with a regression test (using Gatling) against a different cluster.
A major subproject: Developing an end point for importing job templates to Rundeck. A POST request (body containing the job template) to this end point will create a job on Rundeck. Show less

A web scrapper developed in Java to extract the art information of over 300 paintings from a museum website.
Used HTMLUnit to navigate across the website and pull out the information and convert it into JSON.

Dark and deep web are two potential places for the presence of illegal activities (Human Trafficking, selling guns illegally, etc). We used Apache Nutch to crawl for images and web pages of guns, Apache Tika for content and metadata extraction, and Selenium to automate Ajax interaction and more.

Phase 1: A PHP backend program hosted on AWS (Amazon Web Services) connecting to the Zillow REST API for querying property searches. The XML returned by Zillow is converted to JSON which is consumed by one of the following apps.

Phase 2: A responsive web app using jQuery Validations and Bootstrap which requests for property searches connecting to the PHP on the AWS. The requests are made using AJAX to improve the user interactivity.

Phase 3: An Android app again connecting to the… Show more

Phase 1: A PHP backend program hosted on AWS (Amazon Web Services) connecting to the Zillow REST API for querying property searches. The XML returned by Zillow is converted to JSON which is consumed by one of the following apps.

Phase 2: A responsive web app using jQuery Validations and Bootstrap which requests for property searches connecting to the PHP on the AWS. The requests are made using AJAX to improve the user interactivity.

Phase 3: An Android app again connecting to the same backend hosted on AWS to fetch the JSON results. These calls are again made using AsyncTask to improve user interactivity.

Both these apps feature Facebook connections using the FacebookAPI. The property search results can be posted to Facebook as a wall post. Show less

As part of the CS-402 (Operating Systems) course we built parts of kernel for the Weenix operating system.

Phase 1:
Threads and Processes (Single threaded process) - Implementation of PCBs and TCBs
Scheduling and Context switching
Synchronization using mutexes

Phase 2:
Virtual File System
Implementation of various system calls to file system operations.
Testing the VFS with an actual ramfs file system

An AI course project (in Java) to implement logic inference. Given English sentences as the knowledge base, the program can infer (using the backward chaining algorithm) if a given sentence can be proved from this knowledge base or not.

Developed a program in Java which determines the next best move to a given board instance of the Reversi game..
Greedy, Minimax and Alpha Beta pruning were the algorithms used for the implementation.

Emulated a traffic shaper which transmits packets controlled by a token bucket filter using multi-threading within a single process in C.
Also implemented functionality of user interruptions to end the program in a graceful manner bringing the system to a steady state.
After the simulation, the statistics of the packet and token drops are shown.

Maintenance of the DCP oneTouch Application. oneTouch is an application on PhoneGap and SmartMiles which is an umbrella of various approval apps.
Eg: Purchase Orders requested from field engineers could be approved by managers. This app connected to Maximo in the backend to confirm the approvals.

AirWatch administrator for DCP Midstream. AirWatch by VMware is an enterprise solution for content and email management on the mobiles.

Developed the front end for the SALES application - for Larsen and Toubro's Sales team.
Developed a responsive UI using jQueryMobile which could adapt to various screen sizes and resolution.

Developed a mobile app for Leave Approvals by manager. Backend written in Java and the front end in HTML and javascript, it connected to an SAP backend communicating using SOAP web services.
It provided a much needed feature for the managers to approve leaves on the fly.

Developed an app to showcase the Holiday List of the year for Larsen and Toubro. Written in Java on SmartMiles, it connected to the MS-SQL server to fetch the data.

It was a proof of concept proposal to Honda corp.
A full stack development of the QA Management process on Android..
Quality engineer has the ability to take pictures of a defective part in a car and pinpoint it with a little description about it.
A service engineer is notified about it to solve it and send a message back to the QA team to verify it.
It also included report generation feature to tabulate the open and closed issues.

As part of the selection process at L&T Infotech I was part of the team which developed a bank application.on ASP.NET and ADO.NET.
It demonstrated the basic features of a bank. Depositing, withdrawal and transfer of money were the initial ones. As part of the change request, an ATM feature was later added.

This application is an implementation of the IEEE paper - Delay analysis and optimization of scheduling policies in wireless multi‐hop networks
Developing a simulated application in Java for transmission of packets in the shortest delay from one tower to another.
The shortest path was calculated using the Dijkstra's algorithm.

Research on why language processing systems have not met the expectations despite the tremendous advancement in technology.

Inferences:
Underlying emotions, pragmatic interpretations and the methods proposed so far are the reasons for systems not being robust.
If these factors are taken into consideration, then a system can be built that can be of great aid to the differently abled people.

At Ticketmaster, we have numerous internal products. And it is vital to know how much it costs to have them running in the cloud. I implemented the cost expenditure algorithm to first find all the associated resources for the product, and then convert this to an exact dollar amount. This, in turn, helps the management to decide if the ROI is reaching the expectations.

I developed a string matching program using the following algorithms:-
Levenshtein, Needleman-Wunsch, Smith-Waterman, Jaro-Wrinkler, Soundex and TF-IDF.
Currently in my free time I am trying to implement the Jaccard algorithm.

There are 2 files with 4 columns each. Column numbers in the 1st file correspond to the column numbers in the second. All of the string matchers are used to output the best matched string and also the accuracy achieved in the process