Job Requirements
We are seeking an Agentic AI Developer to design, build, and operate autonomous AI agent applications on Amazon Bedrock AgentCore. The role centers on AgentCore Runtime for deploying agents at scale, and on connecting agents to enterprise tools and data through the Model Context Protocol (MCP). The successful candidate will work across multiple workstreams, partnering with product, data, security, and platform teams to move agentic solutions from prototype to production.
Key Responsibilities
Build agentic applications on AgentCore Runtime, including agent orchestration, tool use, memory, and session management.
Design and implement MCP servers and clients to expose enterprise tools, APIs, and data sources to agents in a secure and reusable way.
Integrate agents with foundation models through Amazon Bedrock, selecting models and prompting strategies that balance quality, latency, and cost.
Define agent evaluation approaches, including task success metrics, guardrail checks, and regression testing for non-deterministic behavior.
Implement observability for agent workflows, covering tracing, logging, token and cost tracking, and failure analysis.
Apply identity, authentication, and authorization controls to agent actions and tool access, following least-privilege principles.
Collaborate with application, data, security, and DevOps workstreams to define requirements, deployment pipelines, and operating procedures.
Document agent architectures, MCP tool contracts, runbooks, and known limitations for engineering and business stakeholders.
Required Qualifications
Hands-on experience building LLM-based applications or agents using Python or TypeScript.
Working knowledge of Amazon Bedrock and, preferably, Amazon Bedrock AgentCore (Runtime, Gateway, Memory, or Identity).
Practical experience with the Model Context Protocol (MCP), including building or consuming MCP servers and clients.
Understanding of agent design patterns such as tool calling, planning, routing, and multi-agent orchestration.
Experience deploying containerized services on AWS (for example, using Docker, ECS, EKS, or Lambda).
Familiarity with prompt engineering, retrieval-augmented generation (RAG), and vector search concepts.
Ability to communicate technical trade-offs clearly to both engineering and non-technical stakeholders.
Work Experience
10-15Years
