CodeVerse

CodeVerse is an AI-powered personalized programming layer that helps beginners learn Python through concepts, words, and themes they already understand. It analyzes each learner’s interests, background, and difficult topics, then creates a consistent personalized vocabulary. For example, a space-themed learner might use “launch” for def, “orbit” for for, and “transmit” for print. CodeVerse is not a simple keyword replacer. Personalized code passes through a language pipeline: lexer, parser, Universal Abstract Syntax Layer, Python generation, validation, and sandboxed execution. The platform shows personalized code beside generated Python, helping learners connect their mental model to real programming concepts. AI supports profile interpretation, terminology generation, explanations, and adaptive guidance, while the deterministic compiler guarantees correctness and predictable execution. This hybrid architecture separates creative personalization from safe execution. The platform includes interactive exercises, personalized error messages, progress tracking, and a gradual transition to standard Python. As learners improve, personalized terms are reduced until they can work confidently with real syntax. AMD/Gemma Implementation Status: CodeVerse’s compiler, personalized syntax system, learning modules, and sandboxed execution are fully operational. We developed an AMD ROCm-compatible pipeline for adapting and serving Gemma on AMD Developer Cloud. Due to hackathon time and compute constraints, Gemma fine-tuning was not completed for the live demo. The production compiler remains deterministic and does not depend on LLM-generated code. CodeVerse is designed for students, self-taught developers, educators, bootcamps, and learners who understand logic but struggle with programming terminology. Its goal is not to replace Python, but to create a bridge into it: the environment adapts to the learner first, then guides them toward the real language.