An assistant powered by AI
As an intern at Mad Impact, a Web and Mobile agency based in Villeurbanne, I played an active role in developing an in-house virtual sales assistant project. This project aims to enrich the online shopping experience for e-commerce shops by offering a chatbot that recommends relevant products based on users’ needs. The main aim of this assistant is to increase the conversion rate, improve the user experience and maximise customer satisfaction.
My contribution
My role was to design and develop the assistant’s backend, using technologies such as Flask for the RESTful API, Elasticsearch for data search, and SpaCy for advanced processing (language processing or NLP) of customer queries. I also integrated the OpenAI API to provide personalised and relevant responses to users, with a focus on reducing costs and optimising performance.
The benefits of this project
At Mad Impact, I was welcomed into a friendly and collaborative environment, where the dynamic team enabled me to progress thanks to attentive supervision and a culture of initiative valued highly. The enriching human interactions strengthened my professional and personal bonds, creating a memorable experience.
My internship gave me a wide range of technical experience, including data optimisation with SpaCy and Elasticsearch, and back-end development with Flask. I developed a deep understanding of operational efficiency and the importance of documentation and modularity in development projects.
Working at Mad Impact has enabled me to confirm my aspirations to become a data analyst, by applying my theoretical skills to concrete and stimulating projects. The collaborative environment has enhanced my project management and communication skills, preparing me for my future career with confidence.
The future of the assistant
This project has a bright future ahead of it, with impressive potential to become a truly personalised virtual sales assistant, perfectly tailored to the needs of end customers. In the future, the assistant could also be used to gather user opinions, collect data on market trends for certain products, promote specific product selections, and many other possibilities.