Experience with LLM fine-tuning, LoRA/QLoRA, knowledge distillation, quantization, large-scale LLM serving with vLLM/SGLang, RAG/KG-RAG, or multi-agent systems, including reusable Agent Skills. Hands-on experience with LangChain/LangGraph, supported by deep knowledge of agent harness engineering: context and memory management, state persistence and checkpointing, tool execution, retries and recovery, human approval flows, guardrails, tracing, and automated evaluation.