SaCampell

Projects

~/Projects$ cat Tables of Contents

  1. DOTA 2 - Outcome Prediction
  2. ProceZeus - Chatbot & Machine Learning
  3. MDA - Manufacturing Automation
  4. MCGA - Android App
  5. ADS - Hacking & Testing
  6. SAE - Data Acquisition

~/Projects$ cat Dota2.txt

alt text

About
DOTA is an online multiplayer game which stands for Defense of the Ancients. The game is developped and published by valve and represents the second iteration of its kind. Originally, the game was developped by someone with the alias: “IceFrog” who published DOTA through a community-created mod using the Warcraft III engine.

The game is exceptionally deep and is far more complex than what meets the eyes. The purpose of this data analytic project was to determine whether it is possible to predict if a given team would win simply based on their hero draft. The investigation proved that with an extremely small and finite set of 2000 data points, it was possible to obtain a 63% accuracy prediction on the winning team before a game even begins.

Challenges

  • Data Wrangling
  • Creating dataset
  • Finding hidden features

Notable Research & Development

  • Clustering
  • Classification
  • Entity matching

Frameworks & Technology

Reference

Back to top

~/Projects$ cat ProceZeus.txt

alt text

About

ProceZeus, is the first open sourced AI-powered web chatbot that aims to help people with judicial proceedings and inquiries regarding rental housing issues. The application combines the power of machine learning, natural language processing, and analysis of over 40 000 Régie du logement judgements with the goal of helping users understand possible outcomes to disputes they decide to take to court.

Challenges

  • Unlabelled data preprocessing
  • Unsupervised clustering
  • Natural language understanding

Notable Research & Development

  • Clustering: K-means, HdbScan, DbScan, and Affinity Propagation
  • Preprocessing: Autoencoders, tf-idf, sentence vectorization, and regex
  • Classifiers: decision trees, svm, and deep neural networks
  • Regression: svm, and deep neural networks

Frameworks & Technology

Reference

Back to top

~/Projects$ cat MDA.txt

alt text

About

Maxar Technologies Ltd. is an international space technology company. Participated in OneWeb, a contract which envolves the development of a constellation of 648 satellites estimated at 3 billion $, in order to provide the world with global internet broadband service. The development of automation on the manufacturing floor was essential and groundbreaking for the company due to skepticism of machine performance on meticulous tasks. According to Wikipedia, satellites belonging to OneWeb should begin to orbit as early as 2019.

Challenges

  • Developping 3rd party software to bypass hardware functionalities
  • Failproof systems with empirical metrics

Frameworks & Technology

Reference

Back to top

~/Projects$ cat MCGA.txt

alt text

About

MCGA is a mobile Campus Guide application that is meant to help new incoming students navigate and familiarize themselves with Concordia University Montréal’s two campuses. Both indoor and outdoor navigations are supported by this software.

Challenges

  • Binarizing map images
  • Generating walkable paths for indoor navigation
  • Memory allocation – By default everything runs on the main thread

Notable Research & Development

  • Mathematical morphology for image processing
  • Heap, stack, garbage collector, memory allocation on android devices

Frameworks & Technology

Reference

Back to top

~/Projects$ cat ADS.txt

alt text

About

Headquartered in Montreal, Canada, Automotive Data Solutions Inc. is a technology company specialized in the development and marketing of aftermarket automotive electronics. Their areas of expertise include: immobolizer bypass, data bus integration, audio integration, remote start. Helped develop a tool to test the immobolizer bypass as well as the remote start as part of the continuous integration project. This operation was possible through the simulation of every car on the market. The work relied on heavy hours of pattern recognition, decryption and low level forms of communication.

Challenges

  • CAN bus emulation through hardware
  • Hexadecimal pattern recognition
  • Exploit database for every road vehicle
  • Debugging between code, electrical circuit, & hardware capacities

Notable Research & Development

  • The use of FPGA to accurately mock electrical communications

Frameworks & Technology

Reference

Back to top

~/Projects$ cat SAE.txt

alt text

About

Concordia SAE, located in Montreal, Quebec, is a student-run organization of SAE International at Concordia University. A world-renowned engineering organization, SAE International hosts the Collegiate Design Series for students. These yearly international competitions allow students to apply classroom knowledge to real-life situations, challenging students to design, manufacture, test and compete with a vehicle in a dynamic environment. Designed PCB for data collection of the vehicle on the Baja team. Sensors used consisted of: accelerometer, gyroscope, hall effect sensor, shock sensors, steering potentiometer, and IR thermometer.

Frameworks & Technology

Chalenges

  • Hardware system that won’t break given extreme physical conditions
  • Sensor concurrency and race conditions
  • Stable electrical communications despite vibrations

Notable Research & Development

  • Designing circuit boards
  • I2C, SPI, CAN, & UART

Reference

Back to top

~/$ ▮