I build agentic AI and LLM infrastructure — systems that connect models to tools, memory, retrieval, and real-world workflows.
Currently working on CodePilot, an open-source programmable AI runtime for agentic automation with custom tool execution, streaming inference, MCP integration, and protocol-driven runtime control.
- Agentic runtimes and custom tool-calling protocols
- Multi-modal LLM workflows and automation systems
- RAG pipelines, embeddings, and vector retrieval
- MCP clients, tool servers, and orchestration layers
- Async Python backends for AI applications
- Open-weight model workflows and local inference
- Linux-based runtime environments, isolation, and MicroVMs

