Artificial Intelligence

We design and build machine learning and artificial intelligence systems that take advantage of lastest algorithms and pretrained models and are crafted to solve specific business problems. These are some examples from previous projects.

Educational Chatbot

In this project we designed and built a chatbot for an educational company. It is a chatbot that helps students improve the recall of concepts learned in courses.

The chatbot uses a neural network model for natural language understanding and recurrent neural network for dialog state prediction. The bot is integrated with Slack, Facebook Messenger and Microsoft Teams.

Natural language understanding of emails

A lot of valueable information is exchanged through emails, many times sent to hundreds or thousands of receivers. The information is made for human reading and that makes it hard to be consumed and automated. Our project was to build a natural language understanding module capable to classify the content type of the email and extract the list of specific entities into a database that is the base product of a new startup company.

We implemented a hybrid deep model and rule-based system capable to extract over 80% of the entities shared by many different forms of free text, including documents in PDF, Excel and screenshots of text information.

Recommendation System

In this project we developed a new recommendation system for a global retailer of food supplements. The recommendation system uses a predictive model based on client features including features from purchase history and was designed to take advantage of some features specific of this retailer.

The new system achieved better performance and was able to help the retailer increase the sales conversion.