What is shl_assessment_consultant?

SHL Conversational Assessment Recommender is an AI chat system that helps recruiters and hiring managers find the right SHL assessments through natural language conversation. Instead of searching a catalog manually, you describe the role you are hiring for and the system suggests relevant assessments, refining its recommendations as the conversation continues. Under the hood it uses a Retrieval-Augmented Generation (RAG) pipeline, a pattern that combines a database search step with a language model generation step. When you send a message, the system converts it into a numerical representation using Sentence Transformers, then searches a pre-built FAISS index of SHL assessments to find the most relevant candidates. A Groq-powered language model generates a grounded conversational response based on those search results rather than relying on general knowledge. Each response includes structured recommendations in JSON format with assessment names, URLs, and test types. The system supports multi-turn conversations, meaning it tracks context across several exchanges so you can ask follow-up questions or narrow down recommendations. The API is stateless, since each request includes the full conversation history, making it straightforward to deploy. It is written in Python and runs as a FastAPI service deployed on Render. Challenges noted in the README include reducing hallucination, keeping responses grounded to the SHL catalog, and handling cold start delays on free hosting. The project was built as part of the SHL AI Intern Take-Home Assignment.