Louis Millette

I have built multiple web apps in the Django framework, mostly small ones hosted on AWS. Django is not taught at the University of Waterloo, even in the CS department.

• UWtext: This is my first big project, it's a website for aggregating textbook data and linking it to Facebook posts. It is built in Django/postgres, and hosted on AWS (EC2/EBS + RDS). Over the years I added some metrics on the data I gathered, and put it into a dashboard.
• Skimp.ca: This is my most recent project, it's a website that aggregates subletting data on Facebook (I find Facebook a good source of free data) and makes it easy to sift through. Hosted on AWS (EC2/EBS + RDS) with Django and postgres, I used python3.6 + tensor flow + keras/SciPy to build a machine learning model to categorize Facebook posts.
• Fintechinstitute: For my employer at Seed Sustainable, I helped build the website and hosted web server for fintechinstitute, a financial school. I implemented user login/AUTH and user token systems; unfortunately, the project was scrapped and never made it online. I did receive a letter of recommendation from my employer at Seed Sustainable, who is also one of my references.

Most of my web apps are hosted on AWS EC2/EBS instances, and use an RDS instance for the database. I also dabble in lambda instances, when they feel more appropriate. This is the kind of skill they don't teach at school, CS degree or not.

• UWtext: This is my first big project, it's a website for aggregating textbook data and linking it to Facebook posts. It is built in Django/postgres, and hosted on AWS (EC2/EBS + RDS). Over the years I added some metrics on the data I gathered, and put it into a dashboard.
• Skimp.ca: This is my most recent project, it's a website that aggregates subletting data on Facebook (I find Facebook a good source of free data) and makes it easy to sift through. Hosted on AWS (EC2/EBS + RDS) with Django and postgres, I used python3.6 + tensor flow + keras to build a machine learning model to categorize Facebook posts.
• Financial Calculator: My financial calculator uses a little bit of corporate finance and some python to generate real world financial possibilities. Mainly, it answers the question "If I invested x amount of money, adjusted for inflation each year, for y years in fund z, how much money would I have?" After adjusting for inflation, taxes (federal, state, provincial) , dividends, taxes on dividends, compound returns, this problem is more complicated then you might imagine. The code is hosted on AWS (lambda), the visuals are generated with the help of highcharts.

my SciPy experience is mostly centered around combinatorial optimization and modeling and machine learning. While SciPy is not taught at the university of Waterloo, some stat courses (the machine learning one) assume you have taught yourself by the time you take the course.

• Index Funds: This is my investigation into Index Fund Returns; it's built on the same platform as my investment calculator, but takes a more technical look at the last 40 years, fees, and contrasting 'safe investments' to index funds. I used a similar model to the CAPM, adjusting investment strategy to lower risk with age and optimizing with a Bashin-Hop algorithm instead of LaGrange multiplier; to account for the additional complexity of the model including tax, dividends, etc. I used the SciPy-python framework to implement the model, python + numpy to build the compound multiplier, and plotly to visualize it all with heatmaps, bar plots, and the gamma distribution overlay.
• Skimp.ca: This is my most recent project, it's a website that aggregates subletting data on Facebook (I find Facebook a good source of free data) and makes it easy to sift through. Hosted on AWS (EC2/EBS + RDS) with Django and postgres, I used python3.6 + tensor flow + keras/scipy to build a machine learning model to categorize Facebook posts.
• KKbox Challenge: This one's a final project for my machine learning class STAT 441, actually a intro level graduate course. I would've added a link but I was the only one who used git (group projects am I right?) We built a classifier using the LightGBM algorithm (decision tree + boosting), and pythons very fast H5 storage. Ended up getting a 73 on the project, not too bad.

Numpy is in the background of all my math based projects, but specifically I used it in my combinatorics and machine learning classes. Math with matrices in python