More about me
I am a Full Stack Developer with a passion for building applications that are user friendly and easy to use. I love to explore new technologies, libraries and frameworks. Always on a quest to discover more.
Notable Open-Source Projects
It is one of the most awaited Event in D.G Ruparel, attended by more then 2000 students, spanning across different streams and branches. Dot Tech App is the gateway pass to the Fest, it provides a hassle-free platform to Participate in Events. Register once and participate in any event on fly.
XKCD Browser
A simple native android app using MVVM architecture, with various tools recommended by google like Jetpack libs, Live Data, View Models, DataBinding, Glide, Material Components, ROOM persistance, Kodein for DI, Retrofit2+Moshi,
TimerFX
A utility Windows Desktop Application which always stays on top of every windows, at the center-top of the screen. Featuring quick access to the most commonly actions such as, Screenshot snipper, Clipboard-Manager, Reminder Manager,
The Brain Vault Project
Personal Coaching Startup
Frontend
Backend
Testing Libs
The Brainvault is a personal coaching startup with various offerring ranging from special Events, Coaching sessions, to courses and corporate plans.
I built the web app that fullfills all the needs using best technologies. Gatsby is used for client site and React is used for admin. The backend is inspired from serverless architecture, using Firebase Cloud functions. It integrates our Payment gateway, Zoom meets, AWS SES, Authentication . Serverless allowed faster development and a massive gain in infra costs and management.
I've built the admin app in mind to enable live changes on website, which enables creating coaching sessions linked with any number of zoom accounts, full courses, events, on demand personalised discount coupons.
Every page and component is well tested using industry leading testing frameworks, which covers needs from testing form fields and payment gateways to dispatching emails and newsletters.
This was my longest and largest project that i have worked on in 2021. The project took nearly 7 months for completion. In the end, it was well recieved by the founders and is now actively used by the users.
Omega Administration Toolkit
All-in-one management solution
The Toolkit is a solution for managing bussiness, it aids the Omega Dental App . Using same data-sources as online web site and android app, enables the app to propogate changes to all platforms at the same time. It supports multiple regions.
I built the Android app to solve challenges faced by day to day working of business, the biggest challenge was to keep track of offline orders with online order from other sources. This mobile app tries to integrate all platforms under one app. The app is very small in size as it uses Native Android which provides seamless native user experience. The Backened is built on Firebase, heavily using Firestore and Cloud functions. It uses various Java Libraries like iTextPdf .
Apart from usual CRUD bussiness features, the app provides various other features like generating and printing invoices, sharing products to various social medias, generating on demand personalised product catalogues, notifications service to the owner for every action performed by employees, visualize orders over google map, tracking payments.
All this with the support for multi-region. Making it easier to expand business in different regions without the need to update app.
This has been under active development since early 2020, and has registered 1000s of orders amounting to about 10 lac till now (Fall 2021).
Omega Dental App
Ecommerce solution for Dentistry
Omega Dental App is an dental ecommerce app built for Omega Enterprises . Its a fully functional ecommerce app for Dentistry. You can find the app HERE.
The app is built usingNative Android based on material design principles. The backed in built onFirebase.
The app Enables dental professionals to access dental products ranging from a vast variety of categories, such as Endodontics, Dentistry material, Restoratives, Dental Equipments, Consumables, etc. It minimises the time and effort required to purchase products, particularly in dentistry which has enormous range of products from various brands. The UI is designed to be simple and interactive, it used smooth transitions and animations.
The was my first android project, i had start working on this in early 2019. Initially, the project was written in Java, in mid-2020 i migrated the entire project to Kotlin. Which improved overall deveopment time, drastically reduced bugs, also it enabled access to modern development tools like Flows and Coroutines in Kotlin. The App is still actively used by doctors in Mumbai.
Invoice Maker
Image to Invoice using CNN
A machine learning solution to convert well laid out picture of products to a delivery invoice with supplementary features to aid management. This is one of my favorite projects, I have used Deep Convolutional Neural Network Model inspired from GoogLeNet. The project uses Dentistry Products, however the project can be extended to any other sector. Its trained on 14 different products of contrasting visual features.
The program is build usingTensorFlow and OpenCV I built custom dataset tailored for this project using image segmentation and Annotation tool calledVoTT by Microsoft.
I had collected over 100 Images and 10s of videos from smartphone and annotated them using VoTT. Along with this, the training phase would generate many different variations of images using Data Augmentation. So in total i had over 1400 annotated image, which in the world of CNN was very little. Later. The detection process used OpenCV to segment the object using various traditional approaches like Edge Detection, Contour detection to segment the object in image, which are later fed to neural engine as input and a label is expected as output.
Traditionl 'naive' CNN would obviously overfit my data and fail to generalize for new images. I had tried mixing, truncating and other variations of leading CNN models like VGG, ResNet, and other. In the end GoogLeNet yielded the best result, the Inception blocks proposed in GoogLeNet paper, worked wonders for small dataset of mine.
The project explores internships and jobs in the current market. The dataset is built by scraping publically available web pages of leading websites, Internshala and Monster India, as well as merging well known public dataset - stackoverflow developer survey from the years 2015 to 2020. We have performed extensive data exploration, using various data science and data mining techniques.
Objectives, Build a high-quality dataset and develop an efficient data preprocessing pipeline, Discover trends in current internships and job market, Highlight the areas of potential growth and improvement, Understand current demands in the software industry, detect anomalies
Data collection strategy primarily consists of crawling individual pages(using selenium and requests library) at a time, as within the limitations of the respective robot.txt file for the website, and downloading the html page. Strategy is discussed in depth in later sections. We have used Python as our language of choice, following best practices, guidelines and conventions. Jupyter notebook in a virtual environment for local modelling and exploration and Deep Note for Collaborative workspace.
The Project consists of a Tweeter Dashboard Simulation, that is responsible for simulating real world operation. Redis is used as the In-Memory Database, alongside an RDBMS, Postgres. Postgres acts like a primary central database, assuring ACID properties, while REDIS plays the role of storing output of queries. In this way, Redis DB acts like a cache system.
The project also uses a real world Tweets dataset, having 100,000 tweets, this dataset is used to Simulate High Read/Write operations.
All the database operations for Redis/Postgres are tracked in milliseconds and displayed on the console visible on the front-end.