I enhance and build products with deep learning and consult on how to use AI effectively. Available for hire, Vienna and remote.
Swat.io is one of the leading social media management tools in the
German speaking region. As part of their new automation features, I've
developed a deep neural network that can automatically detect and flag
toxic posts, improving the task of moderating social media pages
Developed with PyTorch, the model implements a state-of-the-art NLP architecture to deliver production-ready performance on such challenging data as social media posts.
By automatically assigning social media posts a sentiment, trends can be detected early on, when there is still time to react. For swat.io, one of the leading social media management tools in the German speaking area, I have developed a deep neural network that accurately assigns one of three sentiments to social media posts, despite a very imbalanced trainings data set.
This project combines the MobileNetV2 architecture of Inverted Residual Layers with a dynamically generated Feature Pyramid Network. The network outputs an array of feature maps with high semantic meaning in different resolutions.
In this multivariate data analysis, I explore the relationship between the key features of my Twitter followers.
Be the first to receive my latest content with the ability to opt-out at anytime. I promise to not spam your inbox or share your email with any third parties.