
AI‑Mentored Code Tutor is a next‑generation learning environment that combines a lightweight, real‑time code editor with a multi‑agent AI system built on Google’s Gemini Flash model. • Mentor Agent analyses a user’s problem‑solving approach, auto‑detects skill level, and provides incremental hints without revealing the full solution. • Code Agent runs the submitted code, checks syntax, evaluates correctness, and suggests optimisations, all in under a second thanks to Gemini‑Flash’s low latency. • Evaluation Agent produces a concise performance summary, tracks hint usage, and logs the session for future analytics. The platform is backed by a FastAPI backend and a Next‑JS front‑end, and it integrates Web‑3 primitives: successful completions mint a unique NFT badge and credit the user with $DSA tokens on the 0G Galileo testnet. By leveraging Gemini’s multimodal reasoning, the system can also accept diagrammatic hints (ASCII sketches) and voice‑to‑text queries, demonstrating a true agentic workflow that guides learners from concept to code to on‑chain reward—all in a single, seamless web experience.
19 May 2026

AI‑Mentored Code Tutor is a next‑generation learning environment that combines a lightweight, real‑time code editor with a multi‑agent AI system built on Google’s Gemini Flash model. • Mentor Agent analyses a user’s problem‑solving approach, auto‑detects skill level, and provides incremental hints without revealing the full solution. • Code Agent runs the submitted code, checks syntax, evaluates correctness, and suggests optimisations, all in under a second thanks to Gemini‑Flash’s low latency. • Evaluation Agent produces a concise performance summary, tracks hint usage, and logs the session for future analytics. The platform is backed by a FastAPI backend and a Next‑JS front‑end, and it integrates Web‑3 primitives: successful completions mint a unique NFT badge and credit the user with $DSA tokens on the 0G Galileo testnet. By leveraging Gemini’s multimodal reasoning, the system can also accept diagrammatic hints (ASCII sketches) and voice‑to‑text queries, demonstrating a true agentic workflow that guides learners from concept to code to on‑chain reward—all in a single, seamless web experience.
19 May 2026