Live
Where the CS50P project (the Dialectical Graphing Calculator) was pure Python — a high-level, forgiving language — this project is written predominantly in C (84.7% of the codebase). Writing a CLI interface, managing file I/O, spawning subprocesses, and piping output between processes in C requires understanding memory management, pointers, and system-level behavior that Python entirely abstracts away. The most technically interesting decision here is the bridge design. Rather than pick one language, the project deliberately splits responsibilities between two runtimes: C handles the user interface, session management, and carbon tracking; Python handles the GPT-2 inference via Hugging Face’s Transformers library. What makes this project distinctive isn’t the chatbot functionality itself — GPT-2 wrappers are common beginner-to-intermediate ML projects — but the carbon footprint layer built directly into the interaction loop.
