Embedded LLM

Singapore, Singapore · ≈30 employees · embeddedllm.com

Embedded LLM builds the infrastructure for production AI. We work across inference, agentic systems and governance, helping AI clouds and enterprises turn GPU infrastructure into reliable AI services. That work spans high-performance model serving, governed AI APIs, stateful agent execution and post-training infrastructure — from open-source foundations like vLLM to products including TokenVisor, TokenVisor Spaces and JamAI Base. We believe the next phase of AI isn't about building more capable models. It's about making intelligence operable: reliable, observable, governable and deployable on infrastructure organisations control. We also believe those foundations should be built in the open. Our engineers contribute to the vLLM ecosystem and to projects like Agentic API, helping open inference infrastructure evolve for stateful, tool-using and long-running agents. From GPUs and runtimes to orchestration and enterprise deployment, we build with the ecosystem, not around it. The goal was never more AI. It was governable AI.