Portfolio

Thesis - Accessible Shopping (Ongoing)

Source

Help visually impaired people shop physically in stores.

Tech Used

  • Languages: Dart, Python, Swift
  • Flutter for creating a cross-platform Android, iOS, and Web app.
  • Graphene - Python GraphQL API framework
  • Tensorflow 2.0 for creating a CNN (Convolutional Neural Network) model that recognizes objects and RNN (Recurrent Neural Network) model that allows the user to query data using speech.
  • Looking into Swift for Tensorflow to serve the API & the models.

Walkability Project (Ongoing)

🔒Source | Article

Co-Author of this paper. Using CNN (Convolutional Neural Network) to identify various street features such as Crosswalks, curb cuts, walk signals and, bike paths. This project tracks 500 participants and measures their physical activity and tries to correlate it with the presence or absence of street features.

Tech Used

  • Languages: Python
  • Frameworks: Tensorflow, Tensorflow Object Detection
  • Tools: Google Colab, Google Cloud AI-Platform

Question Answering dataset & model

🔒Source

We compiled an NLP(Natural Language Processing) question answering dataset which has for each question, an image, a paragraph, and four options with the correct answer to that question. Also, we worked on creating an NLP model that successfully predicts the answer to that question using the image, the paragraph, the question, and the four options. Part of a paper that’s still in works.

VSCode Flutter Extension

Marketplace | Source

Wrote an extension in VSCode that helps with creating JSONEncodable classes. It also runs code generation build tools which remove the need to write a lot of boilerplate methods for encoding or decoding JSON data.

Tech Used:

  • Languages: Dart, Typescript
  • Tools: Visual Studio Code

Spotter (Now Yippee)

Website | Android App | iOS App | 🔒Source

Spotter is a social app for travelers. It allows you to log your trips, add your friends to your trips, follow other travelers on their journeys, and have fun along the way. The app had native Android and iOS apps with a web-app written in React and server implemented in Node.js.

My Contribution

  • Founding member of the startup focused on creating a social network for travelers.
  • Lead the development team of 6 that developed the Android, iOS, and Web apps of Spotter.
  • Mentoring developers and personally contributing to the project by developing RESTful and GraphQL APIs with Node.js, Web development using React, iOS development using Swift, and architecting services on AWS.
  • Used Python and OpenCV for detecting faces in the pictures for improving the quality of cropped thumbnails.
  • Consolidated detailed use cases and requirements through regular meetings with the project partners.
  • Established criteria for each milestone in the product roadmap as a means to measure development progress.
  • Wrote automated integration and unit tests for the GraphQL API and, the React web app.
  • Dockerizing the app so that it becomes portable across any cloud services.

Tech Used

  • Languages: Javascript, Typescript, Kotlin, Swift, Python, C#
  • Frameworks: Express.js, React.js, Android SDK, iOS SDK, GraphQL, OpenCV, Jest.js
  • Databases: PostgreSQL, Redis
  • Tools: Git, Docker

Personal Website

Website | Source

Implemented a static website for me using Gatsbyjs framework and hosted it on a dedicated server. The site serves as a single point of reference for my blog posts, projects, portfolio, and other interests.

Tech Used

  • Languages: Javascript
  • Gatsby.js is a static site generator which uses React, Javascript & GraphQL internally.
  • Ubuntu Linux as the operating system to host the website.
  • Caddy as the static file server to serve the site and manage the Let’s Encrypt certificates.

ATMA - ATtendance MAnagement App

🔒Source

Implemented a cross-platform attendance management system using Angular.js. The app used Chrome apps features as a backend to run on Windows, Mac &, Linux. For persistence, the app used the IndexedDB API available in chrome. Bootstrap CSS framework with themes was used to make the app look good.

Tech Used

  • Languages: Javascript, Typescript
  • SPA Framework: Angular.js
  • Chrome apps feature to run it as a cross-platform desktop app.

Arjun’s Last Lesson

🔒Source

Created a fun little platformer with a fictional story based on the characters in the Indian epic “Mahabharat.” In the game, you take the role of Arjun, the legendary archer, as a youth who is still mastering the art of Archery from Guru Dronacharya. The animals in the forest suddenly become aggressive and start terrorizing the villages. Arjun is given the task of finding a solution to the problem, which leads to him battling huge mutated monsters. Arjun ultimately discovers that the villain is humanity itself for destroying and polluting the beautiful natural environment.

Tech Used

  • Languages: Javascript, Typescript
  • Framework: Phaser.js
  • Chrome apps feature to run it as a cross-platform desktop app.

Flutter Bootcamp Portfolio

Source | Certificate

Collection of mobile apps made during Flutter Bootcamp.

Perception Based Guidance

🔒Source | Video

An Android & an iOS mobile app that helps to guide visually impaired people. We implemented three papers of monocular depth while doing this project.

Miscellaneous projects

  • Speed up algorithms on the web using Rust & WebAssembly.
  • Image Similarity & Classification
  • NBA Team Win Prediction