Katkat
Back

Build and design a Mistral RAG agent

Mistral AISoftware Engineer · Technical Screen
HardSoftware EngineerTechnical ScreenML System Design

Last updated

This question evaluates a candidate's skills in building a retrieval-augmented generation (RAG) system, covering document ingestion, chunking, embedding generation, vector indexing, retrieval, LLM API-driven streaming completions, short-term conversation memory, and operational reliability features.

Build and discuss an LLM-powered retrieval-augmented agent that calls the Mistral API. Implement a minimal Python tool that: (

  1. ingests a small local corpus (e.g., Markdown/PDF), chunks text, generates embeddings, and indexes them; (
  2. retrieves top-k passages for a user query; (
  3. composes a prompt and invokes Mistral chat/completions with streaming; (
  4. keeps short-term conversation memory; (
  5. reads the API token from an environment variable; (
  6. handles errors, retries, and rate limits; and (
  7. includes a brief README and smoke tests. Then, explain your system design choices: chunking strategy, embedding model choice, vector index selection, latency and cost budgeting, caching, prompt templating, safety/PII handling, logging/monitoring, offline RAG evaluation, and fallbacks when retrieval fails.