Uber, San Francisco, CA

Aug. – Dec. 2016

• Developed supervised random forest machine learning algorithm with natural language processing and topic modeling using Apache Spark and Scala to predict how to respond to and resolve low risk customer support issues based on categories with >95% accuracy on 75% of ticket volume.
• Helped architect and integrate the natural language processing predictive pipeline with customer support backend system to decrease handle time by 6% saving $4MM annually in support costs.

Uber, San Francisco, CA

Jan. – Apr. 2016

• Built feature in Python and Go to identify a trip having an inefficient route using an internal routing engine, and automatically refund the difference to the rider upon request, saving ~$215,000/month in customer support costs.
• Implemented merge logic to prevent redundant ticket creation using Elasticsearch, reducing the number of duplicate customer support tickets by over 140,000/month and saving an additional ~$266,000/month in customer support costs.
• Improved code coverage from 62% to 83% and increased instrumentation and error logging to 100% of functional calls while refactoring micro-services migrated from the monolithic backend architecture.

IBM, Toronto, ON

May – Aug. 2015

• Developed a responsive desktop, calculation wizard for complex query modelling in C# with a MVVM architecture, reducing user navigation time by 67%.
• Increased query construction speed by 5% on >1 TB database models using the VS profiler to locate performance deficiencies in the execution, and by resolving legacy code issues.
• Refactored and modularized client-side, legacy, C# WPF code base reducing redundancy by 30%.

Kik Interactive, Waterloo, ON

Sept. – Dec. 2014

• Developed a module to identify spam signatures by employing a K-Means Clustering algorithm which analyzed user activity patterns in Java and identified ~70% of all spam behaviour with 90% accuracy.
• Wrote Python scripts to analyze 10000 access logs for real-time aggregated metrics and edge location server hit rates to invalidate the use of Cloud Front (AWS content delivery network) to save costs.
• Created a Git tool using Python to aggregate a list of merged pull requests from JSON logs to optimize deployment time by 10%.
• Programmed a groovy script to batch send retention emails for return path analysis testing.

SAP, Waterloo, ON

Jan. – Apr. 2014

• Developed a 60-page end-to-end web solution using SQL Anywhere for maintaining synchronization of business or inventory operations in remote areas with latent satellite connectivity, reducing page clicks by 50%, and increasing task completion efficiency by 100%.
• Wrote stored procedures and SQL scripts to execute CRUD operations for 100’s of concurrent transactions using joins and derived tables.
• Architected a web interface for maintaining data from a smart green roof network of distributed sensors; aggregating 1000’s of rows of data using SAP UI5 (jQuery) and SAP HANA.
• Utilized the Model-View-Controller (MVC) software design pattern, and complex JSON modelling.
• Practiced an iterative agile development cycle, with routine unit testing and code documentation.

Zynga, Toronto, ON

May – Aug. 2013

• Executed server side configuration, functional, regression, smoke, stress, and performance testing across various form factors on iOS and Android mobile applications, guaranteeing 90% system coverage.
• Used Charles proxy tool to verify metadata logging and usage information, and JIRA for bug reports.

Prefects Charity Organization, Mississauga, ON

Sept. 2010 – June 2012

• Campaigned in the school community and helped collect over $10,000 in funds to raise awareness for organizations such as KIVA, Engineers without Borders, WWF, and Red Cross

• Created a 3D augmented reality app on the Microsoft HoloLens platform using Unity3D and C# to visualize Ikea furniture in their home to evaluate spatial and aesthetic properties of home décor before making a purchase.

• Researched and assessed viability of automatically surfacing estimated time to resolution of support tickets and identified architectural limitations that needed to be prioritized and overcome to make the feature implementable, encouraging the use of web sockets to track ticket resolution and queuing.

• Created a rudimentary library and conducted experiments to classify Uber support tickets into categories using natural language processing and machine learning with Python’s scikit-learn module in 12 hours.

• Built a chat room based messaging app for television show discussions using a MEAN stack and Meteor.js.

• As a research assistant, built an open source android application to notify the user in real-time and locally tabulate instances of leaked sensitive information triggered by nefarious apps on the device.

• Developed and publically hosted a personal website and blog employing a MVC architecture using the CodeIgniter framework.
• Implemented a login system with basic encryption using PHP, and designed an efficient user interface.
• Wrote and integrated a web scraper using Node.js to fetch and display daily historical facts.
• Created a cache-based photo uploading and rating web utility using JavaScript and Underscore.js.

• Used Node.js and JavaScript to create an arcade of classic 2D games (pacman, asteroids, pong, snake, space invaders and arkanoid) within 32 hours using the canvas to serve as an internal application (card) for Kik, hosted on Heroku.

• Developed and launched to the app store a touch based Brick Breaker game using the Android SDK and the Eclipse IDE for Android smartphones.
• Developed graphics using canvas, and implemented obstacle collision with realistic physics, animations and movement.

• Compiled a set of design documents to implement data visualization for a MDS web portal for the open source Sana health platform compatible with various medical data sets and file formats.

• Developed a command-line kernel for an OS in C with dynamic memory management, scheduling and inter-process communication via assembly language interrupt handling, and a system clock.

• Used VHDL (hardware description language) to program a synchronized system of traffic lights for a major four-way road intersection using the Altera Quartus-II FPGA Design Software.

• Created an assembler using the MIPS assembly language in Java
• Created a compiler that translates a subset of C++ (WLPP) to MIPS using machine code optimization.

• Programmed an LPC17xx ARM microprocessor to record and display the reaction time for pressing a button using LEDs.
• Implemented subroutines, input/output interfacing and interrupt handling in assembly language.

• Integrated software and hardware to program Parallax Scribbler 2 robot that runs autonomously
• Used light sensors to traverse predefined paths, and avoid obstacles.

• Created an interactive database storing employee data using a hash table data structure
• Used efficient binary search and sorting algorithms to allow for data input and retrieval
• Designed a user friendly graphic interface for the database for ease of use

• Programmed the classic arcade game Pac-man in Turing
• Implemented game controls, graphics, saved high-scores, and artificial intelligence (AI)

• Built and programmed a circuit for a functional scale model of an elevator using the C language