My Projects
Software & Data Engineering
End-to-end Google Trends data pipeline with dbt, BigQuery, and Streamlit
A BigQuery data warehouse transformation pipeline using dbt and visualisation using Streamlit.
'Would you have survived the Titanic sinking?' - Multiple Python apps
A repository of multiple python apps written in different frameworks serving the same machine learning model trained on Kaggle's Titanic dataset.
Deploying a Lambda Function in AWS using the Serverless Framework
A web scraper written in Python and configured to be deployed to AWS Lambda using the Serverless Framework.
Dash-vocado App
Visit App on PythonAnywhere | Code on Github
A for-fun data visualisation app built with Dash, using data about avocado sales. This was meant to demonstrate browser-based interactive charts using Python Dash framework.
Natural Language Processing
Detecting Lies from Messages in Diplomacy
A truth or lie classification of messages in Diplomacy, an online board game of deception. The model was trained on a dataset from The Qanta Project.
Eurovision-inspired Lyrics Generator with Tensorflow
View in HuggingFace Models | Code on Github
A text-generation model built with Tensorflow and trained on a dataset of Eurovision song lyrics in English.
Sarcasm Detection with Bidirectional LSTM
A classification model built with Tensorflow and trained on sarcasm texts from Tensorflow public datasets.
NLP Modules for Vector Spaces
Modules and Jupyter notebooks for common tasks in Natural Language Processing, like tweet preprocessing, naive bayes classification, word prediction, and common distance functions; based on the NLP Specialisation by DeepLearning.AI.
Image Processing
An Autoencoder Neural Network from Scratch
An autoencoder framework built in Python, featuring modular implementations of various neural network components, as well as a sample autoencoder for images of Nordic runes.
Cats and Dogs Classifier with Transfer Learning from Inception V3
Demonstration of Transfer Learning from a pre-trained classifier model to enhance the performance of a CNN for classifying images of cats and dogs.
Other Machine Learning
Predicting Sun Spots with Deep Neural Networks
A sunspot prediction model made of a combination of convolutional layers and LSTM layers, trained on time series sunspot data from Tensorflow public datasets.
Geospatial Stuff
Geopandas for mapping universities in the UK
Using Geopandas to plot the top universities in the UK, using data from public datasets and scraped university data from UCAS.