chenxi yan

I build AI backend systems for content products, from Agent decision-making and DAG workflow execution to Redis-backed asynchronous scheduling and multi-model serving. At an AI startup, I built a workflow engine with extensible node registration, long-running task state management, and reusable LLM, image, video, and logic capabilities.

I’ve also delivered multi-GPU AI pipelines, real-time AI voice calls, and commercial backend capabilities for a multi-brand AI Companion platform. My work spans Node.js, Python, Redis, MySQL, ComfyUI, GPU inference, and production integrations across LLM, TTS, image, video, memory, payments, and avatar experiences.