I build practical AI infrastructure: Dify plugins, retrieval integrations, model tooling, and the connectors that turn isolated systems into dependable workflows.

• Dify Plugin Ecosystem: Workflow tools that bring language processing, knowledge operations, storage, and external data into Dify.
• Retrieval and Knowledge Workflows: Integrations across datasets, documents, chunks, memory, and production RAG systems.
• Model and Inference Integrations: Provider support and serving improvements for language, speech, vision, and multimodal models.
• Workflow Infrastructure: The SDK, runtime, persistence, and observability work that keeps AI applications reliable beyond the demo.

• Dify: Web and API fixes across chat state, annotation imports, knowledge ingestion, workflow startup latency, and Langfuse reporting.
• Dify Official Plugins: Provider and model updates for SiliconFlow, DeepSeek, Volcengine, Azure OpenAI, and tool-call message handling.
• Dify Plugin SDKs and Plugin Daemon: Runtime improvements around app context, persistence, and storage accounting.
• Xinference: Serving and deployment improvements spanning worker selection, replicas, GPU metrics, OCR, and speech models.
• RAGFlow: Model-provider additions and retrieval-system integration work.

• Lingua: Detect languages in Dify workflows with ISO language-code output and optional language scoping. Dify Marketplace
• RAGFlow: Connect Dify to RAGFlow for datasets, documents, chunks, retrieval, and memory workflows.
• MinIO: Read, upload, list, and inspect objects in MinIO-backed AI workflows.
• Juhe: Access weather, exchange rates, oil and gold prices, and stock-market data through Juhe APIs.
• Knowledge: Bring existing Dify knowledge-base operations into plugin workflows.

• skills: A collection of reusable agent skills for the workflows I use and maintain.

I am focused on practical AI infrastructure: plugin ecosystems, retrieval systems, model integrations, and the engineering work between a convincing demo and a tool people can trust.