The problem
As the extension to real-time speech recognition and synthesis solution, the client requested additional functionality for their mock interviews service. After completing an interview on their platform, users would have the option to interact with an LLM-powered chatbot and ask questions about their performance – identifying where they excelled and where they could improve during the next conversation.
Industry: HR
Technology: LLM, RAG, Cloud
Our solution
We used Pinecone as a vector database to store complete mock interview content (including multiple interview sessions depending on user activity) along with the client’s internal documents containing best practices and interview guidelines and GPT for interaction with both user and the database. When users asked questions about specific parts of their interview, our solution queried the database to identify the relevant interview segment, paired it with applicable best practices when necessary, and provided accurate responses with actionable suggestions for improvement in real interviews.
The impact
The solution enhanced the client’s existing product by adding a personalized feedback layer, delivering automated, tailored guidance to each user based on their specific interview performance.
