Agentic AI Developer

Overview As a Senior Agentic Application Developer, you will architect, build, and operate production-scale, consumer-facing agent systems. You own end-to-end design—from orchestration to evaluation and deployment—driving reliability and operability in real products. You collaborate with product and platform teams to translate complex problems into agent-driven workflows. This role emphasizes robust stateful architectures, safety, and observable performance at scale. You will shape how agents are designed and proven in production, delivering trusted, high-impact solutions. Compensation / Benefits Weekly pay Medical, dental, and vision coverage Voluntary Life and AD&D coverage Paid Training Opportunity for advancement upon performance and availability Responsibilities Architect stateful agent systems using Lang Graph with complex topologies and long-lived state Design strict, versioned schemas with Pydantic to enforce contracts between agents, tools, and downstream systems Build and operate tool-using agents that orchestrate calls across APIs, databases, and internal services Implement robust failure handling including timeouts, recovery, fallbacks, and human-in-the-loop escalation Instrument agent execution with Lang Smith: custom traces, evaluations, and performance benchmarks Optimize latency, cost, and reliability for production-scale agents Lead memory strategy design (short-term, long-term, episodic, externalized) for agents Own deployment pipelines: versioning, rollbacks, safe rollouts Partner with product/platform teams to translate complex user problems into agent-driven workflows Key requirements 6–10+ years of professional software engineering experience Experience building and maintaining production agent systems with real users Lang Graph expertise (orchestration, state persistence, supervisor/worker patterns)Pydantic expertise (advanced validation, custom types, schema evolution, backward compatibility)Lang Smith experience (trace analysis, evaluation datasets, regression detection)Strong Python, including async patterns and concurrency Experience designing idempotent and retry-safe agent actions Proven ability to mitigate LLM failure modes (hallucinations, tool misuse, infinite loops)Experience implementing guardrails, policies, and safety layers in agent systems Deep understanding of observability for AI systems beyond traditional loggingcross-functional collaborationproblem-solving attitudeability to mentor other engineers Lang Graph Pydantic Lang Smith