AI Engineer

• Agentic AI - Customer & Commercial
  The Agentic AI – Customer & Commercial pod builds customer-facing AI experiences that simplify journeys and enable faster, more intuitive interactions across digital channels. As an AI Engineer on this pod, you’ll build and operate agentic features end-to-end—productionizing retrieval pipelines (embeddings, chunking, relevance tuning), implementing task-oriented agents with clear interfaces, fallbacks, and human-in-the-loop checkpoints, and partnering with AgentOps to automate offline/online evaluations across functional, behavioral, safety, latency, and cost dimensions. You’ll help deliver scalable, production-grade AI on AWS/AICP with strong engineering discipline (testing, Git, CI/CD), while supporting guardrails such as Personally Identifiable Information redaction and audit trails to keep AI dependable and safe in real Customer workflows.

• Agentic AI - ETO Products
  The Agentic AI – ETO Products Pod focuses on building practical AI tools that make internal Teams more effective. As an AI Engineer, you’ll get hands‑on quickly, contributing to agentic features using modern frameworks and learning how to design tool‑driven behaviors that operate reliably in production. You’ll help implement foundational retrieval capabilities and support evaluation and quality checks that ensure AI solutions perform as expected. This role offers the opportunity to grow your skills by working on real enterprise workloads where reliability, clarity, and measurable outcomes matter.

• Design, build, and operate task-oriented agents (planner/executor, tool routing, HITL checkpoints, fallbacks, circuit breakers) with clear interfaces and SLOs.
• Design, develop, and productionize retrieval/grounding pipelines (RAG, structured APIs, metadata/knowledge graphs), including chunking, summarization, embeddings lifecycle, and relevance tuning.
• Develop and document modular system/instruction prompts and tool specs; manage versioned prompt libraries, run A/B variants, and document promotion criteria.
• Collaborate with AgentOps to define and implement per-use-case eval suites (functional, behavioral, safety, latency, cost); automate offline/online evals and wire eval gates into CI/CD pipelines.
• Identify and implement PII redaction, policy-as-code checks, guardrails, and compliant audit trails across agents and context pipelines.
• Determine and optimize context size, caching, rerankers, model selection, and tool routing to reduce cost per successful task.
• Prepare and maintain schemas/data contracts, batch/stream ETL for context stores, embedding jobs, and freshness SLAs (with Data Platform).
• Work with teams to integrate runtime frameworks (e.g., LangGraph/AgentCore, Bedrock Agents/Guardrails) and enterprise systems.
• Maintain agent cards, runbooks, and playbooks; contribute templates and internal libraries for repeatable delivery within the Southwest Airlines AI community.
• Mentor junior AI engineers.
• May perform other job duties as directed by Employee's Leaders

• Intermediate knowledge of embeddings, retrievers, and vector DBs (e.g. pgvector, OpenSearch k-NN); relevance tuning and evals.
• Intermediate knowledge of LLMs and agentic infrastructure such as LangGraph, AWS Bedrock (Agents/Guardrails) or other agentic frameworks.
• Intermediate knowledge of the full AWS stack (Bedrock, SageMaker, Lambda, S3, Redshift, VPC) and enterprise data platforms.
• Ability to translate user journeys into measurable agent behaviors; write clear design docs and runbooks.
• Intermediate experience implementing AI observability tooling for tracing and evaluation.
• Ability to balance innovation with stability in mission-critical AI infrastructure.
• Skilled in agile methodologies and iterative delivery of complex systems.
• Ability to demonstrate conflict resolution, prioritization, and consensus-building skills.
• Skilled in Git-based workflows, automated tests (unit/integration/eval).

• Required: Bachelor's Degree in Mathematics, Computer or Data Sciences, Information Technology or similar fields of study; or equivalent advance level experience

Experience

• Required: Intermediate level experience, fully functioning broad knowledge in:

  • Applied artificial intelligence engineering, including the design, development, deployment, and operation of AI‑enabled systems

  • 2-5 years of relevant work-related experience

  • 4+ years in Data Engineering or Backend Engineering shipping production systems; builds end-to-end agentic features for real user journeys.

  • Designs and productionizes retrieval/grounding pipelines (RAG, structured APIs/metadata); hands-on with embeddings, chunking, and relevance tuning.

  • Builds and operates task-oriented agents (tool routing, HITL checkpoints, fallbacks/circuit breakers) with clear interfaces and SLOs.

  • Partners with AgentOps to define eval suites (functional/behavioral/safety/latency/cost) and automate offline/online evals with eval gates in CI/CD.

  • Implements PII redaction/guardrails/audit trails; AWS stack experience (Bedrock/SageMaker/Lambda/S3/VPC) and strong Git/test discipline.

• Preferred: Experience in:
  • Hands-on AI observability tooling for tracing and evaluation; A/B prompt/model variants.

  • Kafka/Flink or event-driven patterns for context updates and freshness SLAs.

  • Experience in regulated/high-reliability environments (e.g., travel) and responsible AI practices.

  • Experience with LangGraph/AgentCore and AWS Bedrock Agents/Guardrails or comparable agentic frameworks.

• Must meet confidentiality expectations as to confidential, proprietary and sensitive Company information
• Ability to work extended hours as needed