Software Engineer Intern
Worked with the chatbot development team in the biggest tech company in Indonesia. Researched various algorithms and frameworks to build an intent classification and conversational model for Tokopedia's customer service chatbot, as well as integrating it to the Tokopedia's web and mobile applications. Along with other intern colleagues, I contribute in creating numerous intent classifier and conversational models using various NLP algorithms, frameworks and techniques. Subsequently, we also validated the models and compared their performances in generalizing with each other, especially with the current model used, which is built on Google's DialogFlow. Ultimately, the team decided to use the models built on Rasa instead of continuing to use Google's DialogFlow models as it is less costly, more adaptable and still perform well. During this internship program, I assisted the maintenance and evaluation of the chatbot system, ultimately improving the default fallback rate, CSAT score and ticket deflection rate (metrics used). The improvement made on these metrics not only increased user satisfaction, but also directly benefits Tokopedia as the workload of human customer service agents are decreased.
