Show HN: Replacing cloud LLM APIs with local, domain-specific models

Most current LLM workflows depend on cloud APIs, which means sending data outside your system. We’ve been working on an alternative: a fully local stack that lets you run and adapt models without relying on external providers. The idea is not just to run models locally, but to make them useful for specific domains (legal, medical, internal knowledge) while keeping them small enough to run on commodity hardware.<p>Current state: 1) local inference engine (GGUF, API-compatible with existing tools) 2) prototype model hub with REST endpoints and model metadata 3) pipeline in progress to adapt general models into domain-specific ones<p>The open question we’re trying to answer is whether this process can be made reproducible, not just one-off fine-tuning. If it works, it could reduce the need for cloud-based AI in many real-world use cases. Repo: <a href="https:&#x2F;&#x2F;github.com&#x2F;eullm&#x2F;eullm" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;eullm&#x2F;eullm</a> Curious to hear