Senior Agentic Harness Engineer

We are looking for a Senior Agentic Harness Engineer to build and operate a production-grade, governed, AWS-native agentic harness for an autonomous software delivery lifecycle. The role will involve building an agentic loop where JIRA tickets trigger epics and stories, which initiate agents for design, coding, review, and testing, followed by human acceptance/rejection and subsequent agent actions. This is a production engineering role, not a prototype or learning opportunity. Candidates should have hands-on experience building and shipping production agentic systems and be comfortable making architectural decisions in an evolving technology environment.

Responsibilities:

• Build and operate agentic workflows covering design, coding, code review, and testing.
• Implement agent orchestration from trigger through execution, output generation, human review, and subsequent agent actions.
• Develop model routing, retry mechanisms, and provider fallback across multiple LLM providers and models.
• Implement business-continuity mechanisms when a specific model or provider becomes unavailable.
• Own token cost management and context-window management at both agent and workflow levels.
• Implement cost controls, execution budgets, and circuit breakers to prevent runaway agent execution.
• Establish observability, monitoring, and alerting across the agent ecosystem.
• Implement monitoring for agent health, quality, and cost using tools such as Langfuse, Datadog, or equivalent platforms.
• Design and implement AI governance and safety mechanisms, including guardrails, deterministic pre/post hooks, kill switches, and audit logging.
• Implement measures for prompt injection prevention and mitigation.
• Integrate the agentic harness with JIRA/Atlassian APIs and other required business systems.
• Ensure agent actions, decisions, outputs, and human overrides are properly tracked.
• Work closely with internal engineering teams and contribute to knowledge transfer through documentation, demonstrations, and technical handover.
• Participate in architecture discussions and constructively challenge technical approaches where appropriate.
• Deliver through staged milestones, including spike, architecture sign-off, build, and pilot.
• Work in an outcome-based environment where technology choices and implementation approaches may evolve during the engagement.

Requirements:

• 3-5+ years of software/platform engineering experience, with meaningful experience specifically in agentic AI or LLM orchestration systems.
• Bachelor's degree in computer science, engineering, or an equivalent combination of education and practical experience.
• 3-5+ years of software/platform engineering experience.
• A meaningful portion of professional experience should specifically involve Agentic AI or LLM orchestration systems.

Must-Have Skills and Experience:

• Hands-on production experience building and deploying Agentic AI systems.
• Strong experience with LangChain, LangGraph, or equivalent agent orchestration frameworks.
• Experience with LLM orchestration and multi-model/provider routing.
• Experience designing retry and fallback mechanisms for LLM-based systems.
• Practical understanding of AI governance and agent safety guardrails.
• Experience with human-in-the-loop approval workflows.
• Practical understanding of prompt injection risks and mitigation techniques.
• Experience managing LLM token costs, context windows, and agent memory at a production scale.
• Strong Python engineering background.
• Strong working experience with AWS environments.
• Experience with distributed-system observability and monitoring.
• Experience with JIRA/Atlassian APIs or similar ticketing/system-of-record integrations.
• Ability to discuss real-world architectural trade-offs based on systems previously built and deployed.
• Preferred / Nice-to-Have Skills: AWS Bedrock / Bedrock AgentCore, Temporal or similar workflow orchestration platforms, LiteLLM or similar model gateways, Langfuse, Datadog.
• Experience with Cursor or other AI-native IDEs.
• Experience with self-hosted/open-weight models such as DeepSeek or GLM.
• Experience working with Claude Code, Claude Cowork, or Claude Skills.
• Experience in financial services or other regulated industries.

The ideal candidate:

• Has already built and shipped production agentic systems.
• Can explain architectural and technical trade-offs based on practical experience.
• Is comfortable working with evolving technologies and ambiguous requirements.
• Has a strong engineering mindset and can contribute independently to architecture and implementation.
• Is willing to challenge ideas constructively and contribute actively to technical discussions.
• Understands that this engagement is focused on working software and measurable outcomes, rather than fixed-scope documentation or checklist-based delivery.