Beauty Buddy

Beauty Buddy is a Personalised and personable introduction to a more holistic beauty regimen. A chatbot built on google dialogflow/conversation tool that learns about its user the more they interact with it. It learns about your hair type, skin type, weather data, and the various products you are using. The user is asked to keep the conversation going in order for it to keep up with your progress, and by taking pictures of the various products you're using to recommend Uniliver products that are in line with the user’s personal skin and hair types. 

In the backend to recognize the Uniliver products, we used built a Tensorflow model on 100 different Unilever products so we could understand which ones the user is using and suggest better options. We needed to get an understanding of products outside of Uniliver so we used google vision API to find the Objects, Logos, and Text on and image if we did not see any Uniliver products and add the products to a “non” Unilever database. We then used this data to recognize the “non” Unilever products and suggest/recommend a Unilever equivalent.

My role in this project evolved throughout the development. I was a part of the pitch and selling-in of the project as a creative technologist by flying into Vietnam with the creative director and some Google Zoo representatives for a 3-day workshop. I planned and prototyped the Machine Learning aspect and then was assigned to another project. Later I was promoted to Technical Director and took back the project to help wrap it up.