GenAI Engineer

Trinity Industries is searching for an GenAI Engineer to join our Service Analytics organization, supporting rail optimization and shipper decisioning solutions. In this role, you will use Claude and modern AI tooling to build data pipelines, accelerate data science work, and ship production AI capabilities on top of our Azure and Databricks platform.

You will sit at the intersection of data engineering and data science. You will be involved in building data pipelines, training and evaluating models, and building LLM-powered systems - but what amplifies this role beyond a standard DS/DE seat is your fluency with Claude as a development partner: Claude Code, custom Skills, sub-agents, hooks, and the Model Context Protocol (MCP). You will partner with data engineers, data scientists, analysts, and business stakeholders to turn telematics, maintenance, and operations data into decisions our customers and operators can act on.

Join our team today and be a part of Delivering Goods for the Good of All!

What you'll do:

• Design, develop, and operate data pipelines on Databricks ( PySpark, SQL, Python)

• Build and ship LLM applications and agents using the Claude API - document extraction, RAG over maintenance and tariff data, internal copilots, and workflow automation

• Use Claude Code as a primary engineering tool: author and maintain Claude Code Skills (packaged slash-command workflows), sub-agents, hooks, and MCP integrations that let the team build pipelines and analytical assets faster

• Partner with data scientists on model development - feature pipelines, evaluation harnesses, training runs, and the path from notebook to production

• Process and optimize large-scale datasets, including IoT, telematics, and geospatial data, to support analytical and operational use cases

• Establish and enforce engineering hygiene around AI work - prompt evaluation, cost governance (prompt caching, model routing), monitoring, and drift detection

• Translate ambiguous business problems from rail, service, and operations stakeholders into shipped pipelines, models, or AI tools

• Apply version control and collaborative development practices across Azure DevOps repos and pipelines to ensure code quality and deployment readiness

• Identify and implement process improvements and automation to improve pipeline efficiency, reliability, and maintainability

• Partner with management to prioritize data initiatives and align engineering solutions with organizational information needs

What you'll need:

The core test for this role is simple: can you manage pipelines and extract data from Databricks, and can you use Claude as a real engineering partner? If yes, you can do this job.

• Bachelor's Degree Computer Science, Information Management, or related field required; Masters preferred
• 8+ years in data engineering including prior experience in data transformation
• Databricks: hands-on experience building, running, and debugging data pipelines using medallion architecture (bronze / silver / gold). Comfortable extracting data, writing PySpark, SQL, and Python, and managing jobs end to end
• IDE fluency: daily driver in VS Code, Cursor, JetBrains, or equivalent - comfortable in repos, terminals, and modern dev workflows
• Claude Code: hands-on experience with the Claude Code architecture - Skills, sub-agents, hooks, MCP servers, and settings - and how to compose them into reliable engineering workflows. Bring examples of what you have built
• Claude API (or equivalent): production experience with prompt design, tool use, structured output, and evaluation
• Applied data science: comfortable with the model lifecycle - feature engineering, evaluation, and the path from notebook to production (You do not need to be a research scientist)
• Team engineering hygiene: Git, code review, CI, and Azure DevOps repos and pipelines
• Communication: able to explain a model, a pipeline, or a trade-off to a non-technical stakeholder without losing them

THE FOLLOWING MUST ATTACHED TO YOUR APPLICATION:

Submit your resume along with a link to a personal GitHub repository (or public gist) showcasing your Claude Code architecture - Skills, sub-agents, hooks, MCP servers, settings, or any combination you have built. A short README explaining what each piece does and why you built it is more valuable than volume. If the repo is private, a redacted tree, settings.json, and one or two example Skill files are sufficient.

Candidates who include a working Claude Code repo with their application will be prioritized for the technical round.