Developed a robust model size validation feature for the AffineFoundation/affine repository, focusing on enhancing the integrity of the Miner Validation Pipeline. The solution introduced a step that blocks models exceeding 120 billion parameters by estimating parameter counts directly from config.json architecture fields, ensuring reliable and manipulation-resistant validation. Designed to be quantization-proof and fail-closed, the approach defends against malformed or oversized models, reducing deployment risks and improving pipeline performance. The work leveraged Python for backend development and asynchronous programming, with careful attention to data validation and repository hygiene, including documentation updates and a correction to the GitHub URL in TOML.