Senior AWS Agentcore Platform Engineer

Role:

Senior AWS Agentcore Platform Engineer (

Client:

)

Position Type:

Contract to hire after initial 6 months on C2

H Location:

Reading, PA or Exton, PA (Hybrid 2-3 days a week from office)

Job Description:

1.

Observability & Distributed Tracing Gap Analysis:

Assess AWS CloudWatch, X-Ray, Bedrock logging, and AgentCore traces against agentic workflow requirements; produce a comprehensive gap analysis and lead the setup of observability within Dynatrace.

Validation Pipelines:

Design and implement post-deployment validation pipelines for agents and Model Context Protocol (MCP) servers, ensuring deployment health and successful tool registration.

Tracing & Logging:

Implement distributed tracing and structured logging to capture LLM decision logic, tool selections, sub-agent calls, and MCP interactions.

Architecture Strategy:

Evaluate LangFuse and LiteLLM proxies against AWS-native solutions; deliver a target-state observability architecture recommendation. 2. Cost Tracking & TCO (Total Cost of Ownership)

Taxonomy Expansion:

Extend tagging taxonomy to capture costs across agent runtimes, MCP servers, vector databases, and Bedrock token consumption per namespace.

Cost Modeling:

Design a granular cost visibility model to aggregate expenses for agents, MCPs, and LLM tokens by team and department.

Dashboards & Alerting:

Build CloudWatch (or equivalent) dashboards for per-team spending; configure AWS Budgets with proactive alerting thresholds.

Automation:

Automate cost reporting via email and Microsoft Teams, incorporating anomaly detection rules to identify spend spikes. 3.

Monitoring & Incident Management Alerting Framework:

Define and implement P1 P4 alerting rules covering deployment failures, runtime errors, tool invocation failures, and MCP connectivity issues.

Incident Integration:

Integrate alert notifications with Microsoft Teams and email, utilizing resource ownership tags for intelligent routing.

Operational Excellence:

Author detailed runbooks for every alert; publish and maintain these in Confluence to facilitate developer self-service resolution.

Stack Evaluation:

Compare AWS-native vs. third-party monitoring stacks to deliver a long-term recommendation aligned with the broader observability architecture. 4.

Security & Governance Risk Assessment:

Evaluate current IAM and tagging strategies for multi-team isolation; identify scalability gaps and potential security risks.

Policy Engines:

Assess the Cedar policy engine (AgentCore) for fine-grained tool access control and document gaps for enterprise-scale deployment.

Identity Architecture:

Design a scalable Attribute-Based Access Control (ABAC) identity model to ensure multi-team isolation without IAM policy sprawl; deliver production-ready Terraform modules.