Management, UI/UX, Data analysis, Machine learning
Personix helps innovators understand their customers. It's a B2B software as a service where you can drag and drop recordings of conversations and it will automatically highlight and prioritize pains points and motivations. Personix uses state-of-the-art deep learning models to transcribe, analyze and extract valuable insights so innovators can build products people actually want.
We interviewed more than 100 users! At Personix, conducting user research interviews had two main purposes. At the same time that we were collecting customer insights, we were also recording the conversations to later use them as training data for our model.
At the beginning of Personix, I was in charge of product design and prototyping. To build the clickable prototype, I used Adobe XD and applied design thinking principles to iteratively improve the user experience.
Prior to the development of the web app, I was also responsible for carrying out A/B and usability tests with potential clients, receiving feedback, and improving the prototype.
After the research and design stage, I assumed the position of CTO where I had the mission of developing the secret sauce of Personix. The model behind the software should be able to receive conversations in any format, be it video, audio or text, analyze them and extract portions where the user expresses needs or motivations.
Natural Language Processing
Together with a team of 3 developers, in 4 months we collected data, labeled it, and trained a deep learning model that, after several iterations and parameter tuning, managed to achieve an accuracy of 87% identifying pain points and jobs-to-be-done in regular conversations. I was personally in charge of designing the base framework of the model, training, and deploying the system for operations.