Hi HN. I'm an independent researcher and I've been working (completely self-funded) on my inference engine since January 2025. It's built from scratch and uses highly optimized fused kernels and I've been testing it against vLLM and SGLang. Thank you, Victor