Senior Agentic Harness Engineer

We are looking for a Senior Agentic Harness Engineer to work on a production-grade AI-DLC agent harness that enables an autonomous software delivery loop. The role involves building and operating an agentic system where JIRA tickets trigger workflows involving design, coding, review, and testing agents, with human approval and complete auditability. This is a production engineering engagement rather than a prototype or learning-oriented role. The ideal candidate will have hands-on experience building and shipping production Agentic AI / LLM orchestration systems.

Responsibilities:

• Build and operate the complete agentic loop: trigger orchestration agent execution, JIRA output, human accept/reject, and next agent.
• Develop orchestration workflows across design, coding, review, and testing agents.
• Implement model routing, retry mechanisms, and provider-level fallback strategies.
• Work with multiple LLM providers/models, including Claude through AWS Bedrock and alternative/self-hosted models.
• Implement token-cost controls and context-window management at agent and run levels.
• Build circuit breakers and controls to prevent runaway agent execution.
• Establish monitoring, observability, alerting, and health tracking across the agent fleet.
• Work with tools such as Langfuse, Datadog, or equivalent observability platforms.
• Implement agent governance and safety guardrails, including deterministic pre/post hooks and kill switches.
• Implement prompt injection prevention and mitigation.
• Maintain comprehensive audit logging for agent actions, decisions, and human overrides.
• Integrate the agent harness with JIRA/Atlassian APIs and other required business systems.
• Collaborate directly with engineering teams and contribute to architecture and design decisions.
• Document, demonstrate, and transfer capabilities to the internal engineering team.
• Work through staged delivery milestones, including spike, architecture sign-off, build, and pilot.

Requirements:

• Has already built and shipped production Agentic AI systems.
• Can explain technical trade-offs based on practical implementation experience.
• Is comfortable working in an evolving and ambiguous technology environment.
• Can independently contribute to architecture and engineering decisions.
• Communicates clearly and challenges ideas constructively.
• Is comfortable working closely with client/internal engineering teams.
• Understands that the engagement is build-as-we-learn, with milestones tied to working software and outcomes.

Must-Have Skills and Experience:

• 7+ years of software/platform engineering experience.
• Strong hands-on experience building production Agentic AI / LLM-based systems.
• Practical experience with LangChain, LangGraph, or equivalent agent orchestration frameworks.
• Strong understanding of LLM orchestration and model routing.
• Experience designing retry and fallback mechanisms across multiple models/providers.
• Experience with AI governance, guardrails, human-in-the-loop workflows, kill switches, and audit trails.
• Practical understanding of prompt injection risks and mitigation techniques.
• Experience with token cost management and context-window/memory handling at production scale.
• Strong Python engineering background.
• Experience working in AWS environments.
• Experience with observability/monitoring for distributed or agentic systems.
• Experience integrating with JIRA/Atlassian APIs or similar ticketing systems.

Good to Have:

• AWS Bedrock / AgentCore experience.
• Experience with temporal or similar workflow orchestration platforms.
• Experience with LiteLLM-style model gateways.
• Production exposure to Cursor or other AI-native IDEs.
• Experience with self-hosted/open-weight models such as DeepSeek or GLM.
• Experience in financial services or other regulated industries.
• Familiarity with Claude Code, Claude Cowork, or Claude Skills.