Agentic Automation Engineer

Infocap develops human-centric automation for government and regulated organizations. The Agentic Automation Engineer turns business requirements into secure, reliable AI solutions, owning engineering from architecture and prototyping through deployment, evaluation, operational support, and improvement. The role focuses on Amazon Bedrock AgentCore, AWS Quick, and Amazon Connect while maintaining human accountability and agency control.

Responsibilities:

• Design, build, and operate agents using Amazon Bedrock AgentCore, selecting appropriate Runtime, Gateway, Identity, Memory, and Observability capabilities
• Develop agents that retrieve information, use tools, produce structured outputs, and coordinate multistep workflows through reliable APIs and enterprise integrations
• Configure and integrate AWS Quick capabilities for research, business insights, knowledge access, and workflow automation, with appropriate permissions and output validation
• Implement Amazon Connect conversational and agent-assistance workflows, integrating knowledge sources and backend systems while preserving context during human handoffs
• Connect AI solutions to document-processing, case-management, and business systems using relevant AWS services, such as Lambda, S3, and Step Functions
• Combine AI reasoning with deterministic business rules, explicit approval gates, human review, escalation, and recovery paths for consequential actions
• Apply least-privilege access, secure credential handling, sensitive-data protection, prompt-injection defenses, controlled tool execution, and audit logging
• Build evaluation datasets and track groundedness, task completion, tool correctness, failure rates, latency, cost, and operational outcomes
• Use infrastructure as code, CI/CD, monitoring, and documented procedures to support reliable releases and ongoing operations
• Validate service availability and deployment suitability for each customer environment; explain technical choices, limitations, and tradeoffs clearly