Projects

Data Science & Machine Learning

• The Office Reboot: LLM cut
  May, 2023
  Generate novel dialogue for the Office by fine-tuning a LLM on the scripts from the office on a T4 GPU. Used QLora and Huggingface to speed up fine-tuning and store/serve the model for inference.
• Leveraging NLP on alternate data for trading
  Jan, 2023
  Predict the bearish/bullish trends in stock prices by tracking sentiments for companies on the S&P500 using a deep learning pre-trained Fin-BERT transformer leveraging scraped Google News and Reddit wallstreetbets data.
/

• Deep Learning applications in Sports betting
  December, 2022
  Designed a streaming MLOps on Google Cloud (GCP) using a combination of GH Actions, Docker, Databricks, Airflow, DBT, Pytorch written in SQL, Python, Pyspark, Go, to predict football match outcomes.

Data Science & Economics

• Frictions in Dating Markets
  Apr, 2019
  Investigates whether Tinder usage is a function of certain factors, such as regional sex imbalance, income, marriage rates, and presence of other “dating venues”. We collect a dataset of Tinder users in Canada to estimate this model using JavaScript, creating geographic units for estimation using ArcGIS. Analysis is conducted in STATA.

• Investigations into the App Market: Insights from Google and Apple
  Jan, 2019
  Utilizes App store data to estimate and contrast the different consumers on the Apple and Google App stores. Implemented a Berry, Levinsohn, and Pakes Logit Demand model to estimate the effect on the market share of an App on factors such as size of app, price, rating etc.. Data on users, ratings, size, description, price etc. are gathered using web scraping API’s.

Quantitative History

• Peer Effects in the Canadian Expeditionary Force
  Apr, 2019
  Uses a unique Canadian Expeditionary Force (CEF) WWI data set to test the relation between performance (In the form of Medals and commendations) is linked to increased homogeneity in military units, in characteristics such as religion, ethnicity, etc.. Python was used to scrape and organize the data while STATA was used for regression and diagnostic analysis.

History

• Slaves to their Revolution
  Mar, 2018
  Explaining early Soviet adoption of quasi capitalist structures in light of political, social and economic tensions within the party and country from 1921 to 1928. Furthermore points to the necessity of purges from a Stalinist perspective and traditional Menshevik ideology as roadblocks to command economy re-structuring.

Public Policy

• Inequality in Canada: Origins, Consequences, and Solutions
  Mar, 2018
  Considers the case of increasing in inequality in Canada resulting in the consequence of “Western Alienation” whereby western provinces have felt short changed by their eastern counterparts. Furthermore looks at whether this rise in inequality has brought economic benefits and what can be done to ensure more social mobility without raising taxes.