Show HN: Natural language semiconductor geometry generator powered by LLMs (github.com)

A new natural language semiconductor geometry generator, termed GeoForge, has been launched, allowing users to create complex semiconductor designs simply by describing them in English. Powered by large language models (LLMs) like Ollama, Gemini, and OpenAI, GeoForge streamlines the design process by transforming user prompts into detailed geometry specifications. The innovative pipeline features stages for specification extraction, code generation, validation, and export, producing outputs in multiple formats, including GDSII and OASIS, essential for semiconductor design. This development is significant for the AI and ML community as it represents a substantial leap towards integrating natural language processing capabilities into technical design workflows. The smart retry logic and sandboxed validation ensure high accuracy and robustness in generation, while the support for multiple LLMs gives users flexibility in selecting a preferred model. With its user-friendly web interface and comprehensive logging features, GeoForge not only enhances productivity for semiconductor engineers but also opens opportunities for further advancements in AI-assisted engineering, marking a transformative step in the intersection of AI and semiconductor technology.