senior-prompt-engineer

Prompt engineering patterns like Zero-shot, Few-shot, ReAct, and Chain-of-Thought optimize AI agents by improving accuracy and consistency across complex tool and data source orchestration workflows.

Evaluate LLM outputs using dedicated evaluation frameworks and metrics that measure quality, faithfulness, and safety, ensuring your prompt engineering meets production-grade AI system requirements.

Design agent architectures by applying orchestration patterns like ReAct and Plan-Execute to structured output schemes, yielding deterministic results validated through reproducible prompts and evaluation criteria.

Optimize prompts for LLM workflows by running an analysis pipeline that generates optimized versions and validates results, leveraging proven prompt engineering patterns to reduce guesswork.

Multi-agent orchestration patterns enable reliable RAG workflows by structuring agent architectures across tools and data sources, providing reproducible prompts suitable for production-grade AI systems.

Few-shot prompting may not suit AI agent design when deterministic structured outputs are required without examples, as it increases token usage and may conflict with structured output schemes.