Design and maintain ETL pipelines that process and classify unstructured data for Retrieval-Augmented Generation (RAG) systems.
Support the development of agent-based architectures using reasoning and acting patterns such as ReAct.
Build and maintain agent workflows using node-based orchestration frameworks such as LangGraph.
Implement agent memory strategies, including short-term event memory and long-term memory approaches such as summarization, semantic memory, episodic memory, and user preference storage.
Develop system prompts and intent-handling prompts that support reliable agent interactions.
Create evaluation tests and performance benchmarks to measure LLM agent performance.
Build tools that allow LLM agents to interact with external systems and services.
Apply best practices around prompt security, input sanitization, and safe handling of user-generated content.
Requirements
Experience building RAG pipelines or ETL workflows for unstructured documents.
Experience working with LLM-based systems or AI-powered applications.
Familiarity with agent architectures such as ReAct.
Experience building or maintaining workflow orchestration systems (e.g., LangGraph or similar node-based frameworks).
Experience writing system prompts or designing prompt interactions for LLM applications.
Experience evaluating or testing LLM systems.
Understanding of data pipelines and document processing for AI systems.
Familiarity with AWS environments and tools such as AWS CLI or STS.
Understanding of security considerations in LLM systems, including prompt injection and input sanitization.