AI Quiz Builder & Grader
Budget: $250 – $750 USD
I need a Retrieval-Augmented Generation (RAG) module that plugs into my current site entirely through API integration. The goal is two-fold: first, the model should act as a study companion, surfacing context-aware explanations or hints when a student is stuck; second, it must generate and grade quizzes on demand.
The quiz engine has to support multiple-choice, fill-in-the-blank, true/false, and essay or other open-ended prompts. Once a student submits answers, the system should score everything automatically and return item-level feedback in real time. For open-ended responses, I still expect an automated rubric-based score—no manual teacher review. The quiz questions should focus on the aspects that the student need, so for different students, the quiz will have questions focused on different parts of the material.
Please expose clear REST or GraphQL endpoints so my front-end can:
• request a new quiz by topic, difficulty, and length
• fetch generated questions with associated metadata (correct answers, distractors, accepted keywords, etc.)
• submit student responses and receive graded results plus explanations
• access a conversational helper endpoint that uses the same vector store to pull context and craft tailored guidance
I am already running a Python backend, so a solution built with FastAPI, LangChain, and a vector database such as Pinecone or FAISS will slot in neatly, but I’m open to alternative stacks if they meet the same performance and latency targets.
Acceptance criteria:
• questions and feedback are factually consistent with the provided corpus
• average response time under 1.5 s for grade-level quizzes of 10 items
• API documented with Swagger / OpenAPI and a short usage guide
If you’ve previously fine-tuned LLMs for education or deployed RAG pipelines in production, let’s make this happen.
The quiz engine has to support multiple-choice, fill-in-the-blank, true/false, and essay or other open-ended prompts. Once a student submits answers, the system should score everything automatically and return item-level feedback in real time. For open-ended responses, I still expect an automated rubric-based score—no manual teacher review. The quiz questions should focus on the aspects that the student need, so for different students, the quiz will have questions focused on different parts of the material.
Please expose clear REST or GraphQL endpoints so my front-end can:
• request a new quiz by topic, difficulty, and length
• fetch generated questions with associated metadata (correct answers, distractors, accepted keywords, etc.)
• submit student responses and receive graded results plus explanations
• access a conversational helper endpoint that uses the same vector store to pull context and craft tailored guidance
I am already running a Python backend, so a solution built with FastAPI, LangChain, and a vector database such as Pinecone or FAISS will slot in neatly, but I’m open to alternative stacks if they meet the same performance and latency targets.
Acceptance criteria:
• questions and feedback are factually consistent with the provided corpus
• average response time under 1.5 s for grade-level quizzes of 10 items
• API documented with Swagger / OpenAPI and a short usage guide
If you’ve previously fine-tuned LLMs for education or deployed RAG pipelines in production, let’s make this happen.