Senior Software Engineer AI

Senior AI Engineer role on Thomson Reuters’ agent platform team, building and deploying AI agent systems end-to-end using AI-native engineering practices.

Responsibilities

• Design and deliver complex AI agent systems end-to-end, including multi-component pipelines, tool integrations, and multi-step reasoning for initiatives you own
• Build production AI agent systems using modern LLM APIs, agent SDKs, and orchestration/runtime infrastructure; integrate cleanly with existing platform services
• Advance agent observability and evaluation through tracing, automated evals, and quality or drift detection to catch regressions before they reach customers
• Apply and extend progressive delivery practices such as feature-flagged rollout and earlier-stage quality signals to improve release velocity and stability
• Participate in cross-cutting design discussions and technical decisions via RFCs/ADRs, ensuring new systems fit into the overall architecture
• Mentor engineers and help build technical culture by modeling effective AI-native, agentic development practices
• Collaborate with product and engineering peers to translate requirements into solid technical designs

Requirements

• Proven experience architecting and shipping production software systems, with hands-on work building AI/LLM-powered products or agentic systems (not limited to classical ML/data science)
• Direct experience building with modern LLM APIs, agent frameworks/SDKs, or agent-to-tool integration protocols
• Ability to own complex technical projects end-to-end, from design through production delivery
• Strong backend engineering foundation: distributed systems design, service architecture, and cloud infrastructure
• Experience with production reliability and observability: monitoring, evaluation pipelines, incident response, and quality regression detection
• Ability to mentor other engineers and contribute to a strong technical culture
• Comfort operating with high autonomy on a senior-heavy, fast-moving team using AI-native/agentic engineering practices

Technologies

• LLM APIs
• Agent SDKs
• Orchestration/runtime infrastructure
• Feature flagging
• Tracing
• Automated evals
• Monitoring
• AWS Bedrock AgentCore

Location and Salary

• Location: Eagan, MN (hybrid)
• Salary range: USD 100,000 - 204,200 per year

Benefits

• Hybrid Work Model
• Flex My Way: flexible hybrid working, work from anywhere for up to 8 weeks per year
• Grow My Way: programming and skills-first approach
• Comprehensive benefit plans including flexible vacation, two company-wide Mental Health Days off, access to the Headspace app, retirement savings, tuition reimbursement, employee incentive programs, and resources for mental, physical, and financial wellbeing
• Social Impact Institute: two paid volunteer days off annually, plus opportunities for pro-bono consulting and ESG initiatives
• US benefits package including market competitive health, dental, vision, disability, and life insurance; competitive 401k plan with company match
• Paid time off including paid holidays (including two company mental health days off), parental leave, sabbatical leave
• Optional hospital, accident and sickness insurance paid 100% by the employee
• Optional life and AD&D insurance paid 100% by the employee
• Flexible Spending and Health Savings Accounts
• Fitness reimbursement
• Access to Employee Assistance Program
• Group Legal Identity Theft Protection benefit paid 100% by employee
• Access to 529 Plan
• Adoption & Surrogacy Assistance
• Access to Employee Stock Purchase Plan
• Ability to receive an Annual Bonus based on a combination of enterprise and individual performance

Preferred Qualifications

• Experience with developer-facing platforms, plugin/extension systems, or integration ecosystems
• Familiarity with progressive delivery practices (feature flagging, phased rollout, blue/green deployment) at scale
• Experience with managed agent runtime infrastructure (for example, AWS Bedrock AgentCore or similar)
• Domain expertise in legal, regulated, professional-services, or other fiduciary domains is a plus, but not required