Projects & Crafts
Parcelo — Distributed Range-based Job Scheduler
Parcelo is an open-source job scheduler designed for large-scale backfills, ETL workloads, and data migrations. Instead of creating millions of individual jobs, Parcelo treats workloads as numeric or typed ranges and recursively splits them into executable subranges using a binary tree. This enables intelligent pruning, resumability, and distributed execution with minimal orchestration overhead. Read more about it: https://harshmange.in/blog/parcelo-distributed-job-scheduler
Engineered a scalable web-based coding platform allowing users to create, code, and deploy projects with real-time terminal access, capable of supporting 1,000+ simultaneous users without performance degradation. Leveraged Kubernetes for elastic scaling, ensuring efficient resource management and high availability.
Writopedia is an online blogging platform developed in 'MERN' stack that helps users to write about different topics. Programmed token-based email verification system using REST API & MongoDB.
Developed the 'Remote Talk' web-app that provides one click instant meeting without signing up or any payment. Programmed functionalities like video conferencing, live whiteboard, chat and live editor.
Engineered backend of the web app that helps users to schedule the tasks. Programmed scheduling algorithm that generates effective sessions based on some significant attributes. Applied the ‘RAD' model & other Software engineering processes in a team of 8 people.
NodeJS
ExpressJs
MongoDB
React
JavaScript
HTML
CSS
Created a fully functional android app that helps users to manage their time by scheduling daily tasks, Boosted productivity by organizing tasks better using lists, sub-tasks & tags, Tested on more than 70 android devices of different versions, including virtual & physical.
Java
MySQL
Firebase
Developed & Trained a machine learning model from scratch using Logistic Regression to predict the winner of IPL (Indian Premier League) match. IPL is one of the most famous & biggest t20 cricket league in the world. The accuracy of our ML model was around 60-62%.
Python