Lead, AI Engineer
Job Description
Lead Gen AI engineering across Royal Caribbean Group by setting standards, building evaluation and delivery frameworks, and guiding teams on agentic coding and LLMOps.
Responsibilities
- Define engineering standards for Gen AI across Royal Caribbean Group and lead delivery of projects spanning multiple brands and business functions
- Join the GenAI Solutions Engineering team within AI & Analytics and partner closely with the GenAI Center of Excellence (CoE)
- Work with business users to understand requirements and rapidly build prototypes for feedback and iteration
- Act as an advanced daily user of agentic coding tools such as Claude Code, GitHub Copilot, Cursor, and similar; establish enterprise practices for context engineering, instruction files (CLAUDEmd/AGENTSmd), plan-then-implement workflows, subagents, parallel sessions, hooks, and permission/sandbox settings
- Design, build, and maintain skills plugins, slash commands, subagents, hooks, and MCP servers that encode standards and workflows (for example scaffolding ticket creation, PR review, test and eval generation) and distribute via a versioned shared plugin catalog
- Publish living, agent-readable engineering standards that remain short, tested, and current as tools and models evolve
- Define an AI-assisted delivery lifecycle: spec-driven development, small reviewable changes, automated AI-assisted code review, and clear human accountability with guardrails for security, secrets, data classification, and IP
- Establish reference architectures, project templates, and reusable accelerators that are agent-ready and provide a secure, tested, production-ready baseline
- Measure impact of AI-assisted engineering (cycle time, rework, defect rates, token cost) and use evidence to refine prompts and practices
- Partner with the GenAI CoE to publish standards, run design reviews, and retire patterns that no longer support enterprise needs
- Set the testing strategy for Gen AI systems, including unit, integration, contract, end-to-end regression, and security testing for APIs, retrieval pipelines, agents, and tool integrations
- Use coding agents to generate and maintain tests at scale with human-reviewed intent and checks such as mutation testing to verify test effectiveness
- Build test harnesses for non-deterministic behavior using mocked model/tool responses, recorded fixtures, and deterministic seams around LLM calls
- Create verification loops so agents check their own work (tests, type checks, linters, evals) before human review
- Enforce automated quality gates in CI/CD including tests, static analysis, dependency and secret scanning, AI-assisted review, and evaluation checks so nothing ships without passing gates
- Lead blameless post-incident reviews and feed findings back into standards, skills, and tests
- Own the enterprise evaluation framework for LLM applications and agents: golden datasets, rubric-based and model-graded evals, human review workflows, and metrics for groundedness, hallucination safety, latency, and cost
- Create evaluation approaches for RAG (retrieval precision/recall and citation accuracy), tool-using agents (tool selection argument correctness, trajectory quality), and multi-agent workflows
- Integrate evals into CI/CD as release gates and into production as continuous monitoring with thresholds, dashboards, and regression alerts
- Define and validate LLM-as-judge methods, including calibration against human labels and bias checks; guide teams on building representative evaluation datasets
- Govern model prompt and configuration changes through versioned, evidence-based evaluation before promotion
- Lead technical design and delivery of Gen AI initiatives across brands, business functions, and data domains, including cross-domain agents and shared enterprise data and knowledge services
- Architect and develop Gen AI applications and agents using Azure AI Foundry, Azure Container Apps, Azure AI Search, and Azure Databricks
- Build full-stack solutions using Python, FastAPI, Nextjs, and PostgreSQL; design agentic workflows using Model Context Protocol (MCP) to connect agents to enterprise APIs, data platforms, and third-party services
- Design and optimize RAG pipelines including hybrid search, semantic ranking, and vector indexing with Azure AI Search
- Support and extend Microsoft Copilot Studio and Microsoft 365 Copilot solutions; advise when low-code vs custom engineering is the right fit
- Run feasibility studies and proofs of concept, then convert successes into supported production-grade services
- Implement secure authentication and authorization using Microsoft Entra ID, MSAL, and service principals; manage secrets with Azure Key Vault
- Manage APIs and microservices with Azure APIM and FastAPI
- Lead CI/CD and infrastructure automation with Azure DevOps and GitHub, including environment promotion, release management, and rollback strategy
- Apply LLMOps practices: prompt and model versioning, deprecation planning, tracing/observability, drift detection, and cost/token monitoring using Azure Monitor and Log Analytics
- Implement AI safety guardrails such as content filtering, prompt injection protection, and output validation aligned with enterprise security, data classification, and responsible AI requirements
- Mentor and coach engineers across teams on working with AI coding agents and raise expectations for design, code quality, testing, and evaluation practice
- Run hands-on enablement (workshops, demos, office hours) to drive adoption of skills and plugins
- Contribute to the AI governance framework to align solutions with enterprise AI principles, compliance needs, and responsible AI practices
- Communicate technical trade-offs, risks, and progress to engineering peers, product owners, and senior leadership
Requirements
- 7+ years of professional software engineering experience
- 2+ years building and shipping Gen AI or LLM-based applications to production
- Bachelor’s degree in Computer Science Engineering (or related field) or equivalent practical experience
- Advanced daily hands-on proficiency with agentic coding tools (for example Claude Code) including context engineering, instruction files, subagents, hooks, and parallel workflows; ability to review and validate AI-generated code
- Demonstrated experience building skills, plugins, slash commands, subagents, or MCP servers adopted by other engineers, distributed via a shared repository or catalog
- Experience defining and driving engineering standards across multiple teams or products, including reusable templates, shared tooling, and agent-readable guidance
- Strong Python (required); TypeScript or another full-stack language fundamentals for API and architecture design and directing AI agents to maintainable outcomes
- Hands-on testing expertise covering unit, integration, contract, and end-to-end testing (for example pytest), including CI-based test automation and AI-assisted test generation/maintenance for non-deterministic components
- Proven experience designing and operating LLM evaluation: golden datasets, metrics, LLM-as-judge, human review, and using evals to gate releases and monitor production quality
- Deep understanding of Gen AI architectures including RAG, tool use, function calling, MCP, and single- and multi-agent orchestration
- Proven expertise with Azure services: Azure AI Foundry, Azure Container Apps, Azure AI Search, Azure Databricks, Azure APIM, Azure Key Vault, and Azure Monitor, including containerization and infrastructure as code
- Experience with Azure DevOps or GitHub for CI/CD, release management, and environment promotion
- Experience with enterprise identity management (Entra ID, OAuth 2.0, service principals), secure coding practices, and responsible AI implementation including secure use of AI coding tools with enterprise data
- Experience with observability and LLMOps: tracing, prompt and model versioning, monitoring, and cost management
- Track record partnering directly with business users to translate requirements into working prototypes across enterprise scale, plus mentoring other engineers
- Excellent problem-solving skills in a fast-paced dynamic environment
- Strong written and verbal communication skills for explaining technical trade-offs and standards to engineers, business partners, and senior leadership
Technologies
- Claude Code, GitHub Copilot, Cursor, CLAUDEmd, AGENTSmd, MCP servers, GitHub
- Azure AI Foundry, Azure Container Apps, Azure AI Search, Azure Databricks
- Python, FastAPI, Nextjs, PostgreSQL
- Model Context Protocol (MCP)
- Microsoft Copilot Studio, Microsoft 365 Copilot
- Microsoft Entra ID, MSAL, Azure Key Vault, Azure APIM, Azure DevOps, Azure Monitor, Log Analytics
- OAuth 2.0, Azure, CI/CD, pytest, TypeScript
- Vector indexing, hybrid search, semantic ranking
- LLM-as-judge, LLMOps
Preferred Qualifications
- Master’s degree in Computer Science Engineering, Data Science, or related field
- Experience with the Claude Agent SDK or similar frameworks; running coding agents headlessly in CI/CD for review, testing, or automation
- Experience operating an internal plugin or skills marketplace with versioning, testing, and adoption measurement
- Experience with Databricks (Unity Catalog, Genie, MCP integrations) and enterprise semantic or data layers
- Familiarity with Microsoft Copilot Studio and Microsoft 365 Copilot extensibility and Semantic Kernel
- Experience with evaluation and observability tooling for LLM applications and agents
- Experience with multi-cloud Gen AI platforms (AWS Bedrock, Google Vertex AI)
- Experience with MCP server design and agent-to-agent interoperability patterns
- Experience with AI safety testing such as red teaming, prompt injection testing, and adversarial evaluation
- Experience establishing or contributing to AI governance or engineering communities of practice in a large enterprise
Location
- Miami, FL (onsite)