Created python scripts that allowed the Nuance research team to view groups of utterances as a dialogues, enabling them to check if their software was behaving as expected over the course of a dialogue. Improved a parser to extract additional research data the team wanted. Wrote a bash script to automate the parsing of new files. Added support for German users to be able to pronounce english letter abbreviations in German and Germanized english Enhanced a web-app's data visualization features to view specific metrics using HTML, CSS, JavaScript, Jinja2, Python and Chart.js.

Built a classifier capable of determining a conversation path based on data from call centres. The classifier used the text from thousands of conversations and a predetermined set of rules to classify a conversations intent. Created a dashboard to compare metrics between different build versions of the same project, allowing my team to view which metrics were suitable and regressing. Integrated support for custom metric baselines for different build versions in order to determine whether a build version is suitable based on client requirements.

The goal of this project is to provide users with more valuable insights into their investment portfolio's performance than brokerages typically give by default. Responsible for end to end development and deployment of the application. Devised suitable data models for investments and implemented custom user models. Wrote backend logic using Python within Django framework. Achieved computation of performance metrics for a user's portfolio. Designed and built frontend using HTML, CSS and Django's template language. Successfully deployed application on DigitalOcean Droplet using a PostgreSQL database within a VPC.

Responsible for the complete development of the application in Kotlin. Followed Android application best practices and used MVVM architectural pattern. Integrated the application with existing Firebase backend. Successfully delivered on all project requirements outlined at the start.

The goal of this project was to develop a pipeline capable of detecting, localizing and tracking individuals walking across a scene from a fixed viewpoint. Played a major role in the data cleanup and setup, and developing the localizer. Played a minor role in creating the classifier capable of detecting person and non-person patches in each frame.

The goal of this project was to better understand the relation between symptom search trends and COVID-19 hospitalizations across several US states. Google’s symptom search trend dataset and their hospitalizations dataset were used as the principal datasets for this project. Was personally responsible for the majority of the data pre-processing and cleaning. Contributed to reducing the data dimensionality using PCA to better visualize it. Also contributed to the creatiion of the classification models.

The purpose of this project was to implement a deep neural network in-order to perform multi-label classification on a modified version of the MNIST dataset. The dataset is made up of sample images which contain 1-5 handwritten digits that can each be classified into 11 classes. The multi-label classification approach was to split the input image into individual digit images, which we then fed into a deep neural network trained on single handwritten digit recognition. The final accuracy obtained on the test data was 99.392%. Primary contribution to the neural network model and parameter tuning.

Worked on a 6 person team to build and program a robot capable of autonomously navigating a predetermined course. Main contribution was to the testing of the robot, but also completed some of the software development and robot design. Performed version control using Git via Github.