Lin Xingbin

At Trane Technologies, I build production Generative AI systems including enterprise agentic RAG assistants, LangGraph Text-to-SQL agents, and Qwen2.5 fine-tuning workflows. I combine Python, FastAPI, Azure OpenAI, LangChain, LangGraph, React, and Azure services to deliver secure conversational applications with evaluation, guardrails, observability, and reliable multi-step workflows.

Previously at Gridsum Holding, I led AI-driven energy optimization platforms on GCP that helped commercial and industrial customers achieve 15–30% reductions in energy consumption. I built forecasting, anomaly detection, IoT ingestion, MLOps, and closed-loop optimization systems across more than 60 production deployments.

My earlier data science and engineering work at Gridrem Smart Energy and Nexant focused on scalable energy data platforms, real-time telemetry pipelines, predictive models, and analytics that supported major energy savings across customer portfolios.

I bring hands-on experience spanning LLM applications, machine learning, cloud-native architecture, data engineering, and production AI operations. I enjoy translating complex operational problems into dependable AI systems that teams can deploy, evaluate, and improve.