For my final year project at UBC, I built a SaaS service which will infer context from some textual or image input in order to better target advertisment parameters.
The service will infer the time window, place, and sentiment from the input and then suggest advertisement parameters.
In order to do so, we leveraged multiple third parties such as Microsoft Azure, Amazon Rekognition, IBM Watson and more. Additionally, we built our own models and use libraries such as Google’s Tensorflow and Syntaxnet to further process the inputs.
The project was deployed using Docker, TravisCI, and Kubernetes.
Unfortunately, the demo is no longer available as we do not have any remainding Azure credits.