AlgoWhisperer: An Open-Source AI that Guides instead of Gives

Описание и применение

AlgoWhisperer – открытый AI‑ментор, который помогает отладить код, выдавая подсказки вместо готовых решений, сохраняя приватность и качество обучения.

Задача

Competitive programmers need a tutor that helps debug code without giving full solutions, preserving learning integrity.

Результат

AlgoWhisperer provides targeted debugging hints without revealing full solutions, enabling private, cost‑free learning for competitive programmers.

Рабочий процесс

• User pastes problem and code into React UI
• Frontend sends context to FastAPI backend
• FastAPI routes request to LangGraph agent
• LangGraph enforces guardrails and forwards to Qwen2.5-Coder-7B-Instruct via Hugging Face Inference API
• Model returns targeted hint instead of full code
• Frontend displays hint to user

Доказательства

We built Algo Whisperer, a strict, Socratic AI programming mentor designed to help developers debug C++ and Python code without ever revealing the final solution.

Источник

When we ask the chat for help, our React frontend sends this context to a FastAPI backend.

Источник

We are using LangGraph to enforce strict guardrails before it hits the open-weight Qwen 2.5 Coder model on Hugging Face.

Источник

The reasoning engine is powered by Qwen/Qwen2.5-Coder-7B-Instruct, accessed via the serverless Hugging Face Inference API.

Источник