SaaS Project: Built a service called Deliver Refunds that automatically recovered refunds for parcels delivered late.
This SaaS product was used by several e-commerce businesses to recover as much as 20% of their overall shipping costs. I wrote Python scripts that used proxies to scrape shipping data. Late deliveries were flagged and automatically filed for refund. Stripe API was used to automatically charge customers based on the refunds collected for them.
Blockchain Project: Built an app to anonymously save encrypted notes on the Ethereum Blockchain.
This project intended to explore value propositions unique to the Blockchain. It was intended to save a note (or your last Will) in an encrypted form on Ethereum - this note would exist on the network forever, and only readable by those possessing the encryption key.
A smart contract stored all user notes. Another on-chain contract tracked the latest version of the contract in use; this provided me the flexibility to issue software updates regardless of the contract immutability.
Applied Machine Learning to find similar verses in the Quran: This app uses a Word2Vec model to find linguistically similar verses in the Qur’an.
Every verse was feature engineered to reduce dimensionality - this work was performed using Python in Jupyter Notebooks. I used Amazon RDS to store the dataset as the resultant dataset was several million rows long. An API was built in PHP to fetch the results. The front-end of the website was built in Svelte and served over Netlify through the project’s GitHub repository.
Single Page Application using Svelte: Built an app called Portfolio Visualizer to visualize stock portfolio performance.
The data is pulled from Yahoo Finance. I couldn’t find an un-throttled free API so I built my own. The API is deployed on Google Cloud Functions to provide a serverless access point.
Single Page Application using Vue.js: Built an app called Stacker Tracker to track bullion asset prices.
The value of mark-to-market value of physical bullion is difficult to track. This app uses a bullion API to update the value of a user’s total holdings. Instead of requiring users to login with their email, the app creates a unique link so the user’s privacy may be protected.
Mobile App: Built an iOS anonymous chat app called Chatr using Swift for the Apple App Store.
I used Python, Swift and Firebase to build Chatr. A primary challenge with public chatrooms is the non-uniform distribution of users in each room; leading to too much activity in some rooms, and too little in others. I used unsupervised machine learning to automatically segment users into chatrooms. This meant the user, instead of trying to find a sufficiently active chat room on their own, could rely on the app to find one for them. This model can also be applied in other use cases to optimize distribution of resources (for example, multi-player games, call centre receptions, etc).
Reverse Commerce: Built a sales cycle and inventory management software for an LCD refurbishing business called Cash4Screens.
My partners and I created a company to buy broken mobile screens and refurbish them for resale purposes. We equipped a warehouse with laboratory equipment, attained the required IP, and hired staff to turn cracked screens into completely new LCD’s. Cell phone shops sold broken screens to us through the commerce platform I built. It was an e-commerce store operating in reverse: we paid our customers for their products.
Once refurbished, we resold the recycled screens back into the market.
Social Network: I created a social network for geeks called Shuzak.
Shuzak was fairly successful; hosting thousands of engineers from around the world. The network was modeled after Orkut, but aimed at technology enthusiasts. It supported LaTeX for rendering math equations, code lints, forums, and chat. Users accumulated karma points for contribution, and redeemed them for advertising on the network. Shuzak ran for several years, and still has a community that lingers on IRC channels.
Built a Shopify app to visualize and map live e-commerce orders.
Built a Chess app in plain JavaScript to explore advanced features of the HTML 5 Drag and Drop API. The app offers class based methods to create moves using Chess notations, flip board, go back in history, etc.
Built an RSS aggregator called FuelFeed used to automate and schedule twitter posts.
Desktop Application: In school I learned about Eigenvectors and their use in ranking search engine results. I used this knowledge to implement the PageRank algorithm in my first commercial grade application. The product iterated over dictionary word combinations to find recently expired domain names that still possessed value in terms of search engine optimization.
Volunteered at Pakistan Science and Engineering Foundation for 3 years to promote higher education. I built and managed a website to help enable collaboration amongst those committed to the propagation of science in Pakistan.