Senior AI & Data Quality Engineer / Architect

ExaTech Inc is hiring a Senior AI & Data Quality Engineer / Architect to take its AI agent from prototype to production and contribute to a multi-agent platform. The role focuses on building and evaluating RAG systems, implementing guardrails, establishing data quality standards, and creating trusted datasets for AI.

Responsibilities:

• Assess the current AI agent and lead the plan to production
• Build and tune RAG pipelines: chunking, embeddings, hybrid search and grounding
• Automate LLM / agent evaluation and regression testing in CI/CD
• Add guardrails, human-in-the-loop review and monitoring
• Challenge the AI team's technical approach; favour robust solutions over workarounds
• Turn SME business rules into testable requirements and evaluation sets
• Set Data Quality standards, profile data for trends and root causes, and build Golden Layer datasets
• Help architect the multi-agent Wizard and shape the Data + AI roadmap

Requirements:

• Hands-on production delivery is required; conceptual or proof-of-concept experience alone is not enough
• A RAG system built and deployed from scratch
• Automated LLM / agent evaluation (e.g. DeepEval, RAGAS, promptfoo) in a release process
• Pro-code agent development in Python (e.g. LangGraph, Semantic Kernel, Google ADK), beyond low-code builders
• Gemini, Azure OpenAI / AI Foundry or similar enterprise LLM platforms
• Guardrails, hallucination controls and AI observability
• Data Quality: profiling, validation, monitoring, Golden Record creation and insight from data
• Strong Python and advanced SQL; cloud data pipelines (e.g. Databricks, BigQuery, Azure)
• Confidence to challenge engineers with evidence; strong stakeholder communication
• Available in EST hours
• SAP S/4HANA data: hands-on work
• leading a full ERP implementation is a strong plus
• Master Data Management (MDM)
• Supply chain, manufacturing or operational data
• Medical Devices / Life Sciences, including regulated data handling
• Enterprise data governance and data catalog tools
• LLMOps / MLOps practices: prompt and model versioning, cost and latency monitoring