Fine-tuned T5 and GPT2 transformer networks to be able to summarize reviews then expand them into poems.
Used Python with Pandas, NumPy and PyTorch to clean and pre-process large text-based datasets.
Created a web-app to host the project at poetify.anvil.app (Only available on request).
Received a 95% on project presentation and 85% on the final report.
Poetify is a machine learning system capable of taking in a review such as a yelp review or an amazon review and outputting a poem related to that review. Ideally, the poem will retain the general topic as is addressed in the review and will keep the same sentiment towards what is being reviewed.
Stored class questions and votes in a database and updated it in real-time using cloud functions.
Managed the look of the front-end using HTML and CSS and used JavaScript to manage the interactivity.
Created the backend using Node.js and Firebase.
Pitched the product to professors and received a job offer to build a new feedback tool for assignments.
The goal of classAsker is to help students post their questions and support each other's questions during a lecture in hopes of increasing student-feedback to the professors. ClassAsker is a web-app that enables students to anonymously post the questions they have during lecture to a list that only their class can see. Students are then able to upvote the questions they want to see answered. The questions with the most votes get sorted to the top of the list in real-time for everyone to see.
Created an API in C++ to interact with the OpenStreetMap database
Created search, route planning and fastest path functionalities using A* algorithm.
Used multithreading to increase performance.
Created Graphics and a User Interface to interact with the various functionalities.
Worked in a team of three and used Git to manage code.
Received a grade of 83 on the project
In a class at the University of Toronto, we had four months to build some type of geographic information system (GIS). When talking to a friend, we discovered that paramedics in Ottawa don't have access to a GIS with path-finding functionality, so we decided to build one for them. The find-path functionality always finds the fastest path in less than half a second in any city and provides visual as well as written directions to the destination.
