senior-prompt-engineer

Prompt optimization involves refining prompt design using advanced patterns like chain-of-thought and few-shot learning to improve LLM output quality, cost efficiency, and reliability across Claude, GPT-4, and related models in production environments.

RAG system optimization combines retrieval strategies, embeddings, and indexing to enhance how language models fetch and use external data, enabling scalable, grounded responses without retraining the underlying model.

Multi-agent orchestration involves designing state management, tool calls, and coordination patterns so multiple LLM agents can work together on complex tasks, requiring careful prompt structuring and workflow definition.

Structured outputs enable reliable, parseable responses from LLMs by enforcing specific formats in prompts and orchestration logic, critical for production-grade tooling interfaces and downstream system integration.

Evaluation pipelines measure prompt effectiveness through systematic testing, observability instrumentation, and monitoring integration with vector databases and embedding models to detect quality degradation in production.

RAG and agent evaluation requires testing retrieval quality, response grounding, tool usage correctness, and end-to-end task success using metrics tied to production requirements and user outcomes.