Projects done

---

Peer to peer multiplayer game

April 2016

Guide: Prof. Vinay Ribeiro, IIT Delhi

• Developed a peer to peer networked multiplayer ping-pong game.
• The game had several graphics options and difficulty levels, and functioned smoothly even when there was high latency at one player's end.

March '16

Guide: Prof. Vinay Ribeiro, IIT Delhi

• Worked in a team of 3 to develop an android application for managing complaints at IIT Delhi.
• The backend was developed using the Codeigniter framework and was exposed to the app via RESTful APIs.
• The app had separate complaint submission and resolution systems for personal, hostel and institute level all provided inside a single application, as well as features such as discussion threads, commenting, notifcations etc.

August 2016

Guide: Prof. Mausam, IIT Delhi

• Worked on a local search approximation solution to the NP-Hard problem of Combinatorial Auction.
• Implemented greedy local search with random restarts to reach increasingly profitable solutions.

Sep-Nov 2016

Guide: Prof. Rahul Garg, IIT Delhi

• Used Convolutional Neural Networks with dropouts in Tensorflow to classify digits with 99.6% accuracy.
• Augmented Data for training using keras library.
• Tried methods like SVM on PCA reduced data to improve accuracy.

Oct-Nov 2016

Guide: Prof. Mausam, IIT Delhi

• Designed a competitive bot for a Strategy Game TAK, using minimax tree search and alpha-beta pruning.
• Used Temporal Difference Learning for assigning weights to features for better heuristic.

January - March 2016

Guide: Prof. S. Arun Kumar, IIT Delhi

• Implemented a compiler for a simple imperative programming language in SML.
• The compiler had a lexer, parser, AST generator, code generator. Basic library functions were also implemented which could be used by the program and would be used for code generation by the compiler.
• The compiler also had a framework for determining error at each stage in compilation.

Sept'15 to Oct'15

Guide: Prof. Amitabha Baghchi, IIT Delhi

• An Engine that maintains an Inverted Page Index of Webpages using Hash-Maps and AVL Trees which allowed to increase the Search Speed of the Queries.
• It gave results in the Order of Relevance of the Queries. The Queries could be single words or compound words.
• Everything was implemented in Java on Eclispe Software.

August'15

Guide: Prof. Huzur Saran, IIT Delhi

• Deployed the Sudoku using Google App Engine on www.sudokuiitd.appspot.com
• It generates Random Sudokus under the Three levels and Solver(using Backtracking Algorithms) can solve any Sudoku with the initial given configuration and also detect if the initial Configuration of the Sudoku is Correct or Incorrect.
• Each Level had a timer with Score,Reset and Solution Capability.

September'15

Guide: Prof. Parag Singla, IIT Delhi

• Implemented both Multinomial and Bernoulli Naive Bayes in python from scratch (Supervised Learning).
• Laplace Smoothing was done in both cases.
• Accuracy of 95% and 85% was achieved on given dataset in case of Multinomial and Bernoulli respectively.

May'15 to July'15

Guide: Prof. Subhashis banerjee, IIT Delhi

• It contained Map of IIT Delhi in Detail and Each Place had its own Page which people can edit and add photos.
• People could also add a new place or route onto the Catalog.
• It was implemented using the usual web languages and also used Google Maps API.

March'15 to April'15

Guide: Prof. Preeti Ranjan Panda, IIT Delhi

• The Striker had variable initial speed and could be launched in any direction,Also Stiker could have Oblique collisions with the coins, friction was also incorporated.
• Technologies used:C++,Glut Library

September'15

Nutanix Hackathon

• Online portal to connect the Lawyers and Clients,Lawyers could make their profile and search for cases whereas Clients could find lawyers and post their cases.
• It used Restful API's for fetching and putting information in the Mysql Database.
• Technologies used:Codeignitor Framework,HTML,CSS,Javascript,PHP,Bootstrap,Jquery.

August - October 2016

Goldman Sachs ABLE challenge

• Used preprocessing techniques such as local adaptive thresholding to convert image (taken programmatically from mobile camera) into a suitably clear binary image so that OCR can be applied.
• De-skewing of image (in 3D) and noise filtering was done to increase the accuracy of the OCR.
• Tesseract open source library was used to perform OCR on preprocessed image.
• Google text to speech API was used to convert the obtained text to audio. Google translate API was also used to provide a choice of language to user.
• This application is being developed as a product in a contest organized by Goldman Sachs to aid visually challenged people. The app can be used to read any label, title newspaper, book etc. in real time.