Our Client is seeking a Senior Agentic Software Engineer  .

Duties Include:

• Design, build, and operate agentic systems that reliably complete real tasks, not just answer questions
• Architect agents that support planning, memory, tool use, and multi-step execution, selecting appropriate patterns (single-agent, multi-agent, workflow-driven, human-in-the-loop) based on problem constraints and risk
• Balance autonomy with control by designing agents that are predictable, debuggable, secure, and aligned with business goals
• Build agents using modern frameworks (e.g., LangGraph, LangChain, Semantic Kernel, AutoGen, or equivalent), implementing structured outputs, tool-calling, reflection, and state management
• Design and implement MCP- and/or RAG-based integrations as first-class mechanisms for how agents access tools, data, and context
• Enforce security, consent, access control, and observability across all agent-tool interactions, partnering with platform teams to establish and evolve MCP integration standards
• Integrate agentic systems into existing services and platforms via APIs and backend services, owning production readiness end-to-end
• Define what -good- looks like for agents using clear metrics (e.g., accuracy, success rate, latency, cost, failure modes), and use those metrics to drive continuous improvement
• Build automated evaluation pipelines (offline tests, synthetic data, regression checks) and instrument agents with tracing, logging, and observability to support debugging and iteration in production
• Design fallback, recovery, and human-escalation mechanisms for failure scenarios, proactively identifying and mitigating failure modes
• Establish architectural standards and best practices for agentic development, raising the technical bar through design reviews, documentation, mentorship, and knowledge sharing