Forward Deployed Engineer – Agent Runtimes & MCP (Python)
Contract position
Remote (PST hours) - Preferred Pheonix
Role Summary:
We are looking for a hands-on Forward Deployed Engineer with strong expertise in Python, LLM agent systems, Anthropic Claude, MCP servers, authentication, and real-time trust/risk integrations.
This role will build the Agent Trust integration into Anthropic's Claude Commerce Agents ecosystem and supporting Claude tooling. The goal is to make the integration easy for merchant engineering teams to adopt with minimal implementation effort.
The ideal candidate is a strong Python engineer who can quickly understand unfamiliar codebases, build secure agent integrations, and translate trust and identity requirements into production-ready software.
Key Responsibilities
Stand up and validate commerce blueprint verticals across multiple Claude runtimes.
Run and maintain test and verification suites.
Maintain fork compatibility with the upstream blueprint.
Build and maintain the AgentTrustMixin integration for storefront and merchant backends.
Implement:
Session-start identity binding
Trust checks during cart and checkout flows
Approve / deny / step-up decision handling
Structured error and tool-result mapping
Per-hook fail-open / fail-closed policies
Decision logging without exposing PII
Outcome reporting
Build and test MCP server integrations for Claude Managed Agents.
Implement authenticated gateway patterns for MCP access.
Support multiple identity mechanisms, including:
Vault credentials
Signed claims
OAuth-based authentication
Build an Experian MCP server for advisory/risk tools.
Develop a companion Claude Code skill/plugin for commerce-builder workflows.
Create merchant integration documentation and implementation guides.
Measure adoption effort using metrics such as:
Number of modified files
Lines of code changed
Prompt bytes added
Number of trust gates
Time to first verified session
Demonstrate and validate the solution against the retail commerce vertical.
Must-Have Skills
5+ years of strong Python development experience.
Advanced Python skills, including:
asyncio
Async programming
Type hints / typed Python
pytest
ruff
Strong object-oriented programming skills.
Experience subclassing abstract classes/backends and extending framework behavior.
Experience mapping exceptions and backend failures into structured API or tool responses.
Hands-on production experience building tool-using LLM agents.
Strong experience with:
Anthropic Claude Messages API
Claude Agent SDK
Understanding of prompt caching and optimization of static prompt prefixes.
Experience building or integrating Model Context Protocol (MCP) servers.
Experience with streamable HTTP MCP servers.
Strong understanding of server-side enforcement for agent/tool-based systems.
Experience securing MCP servers with authentication.
Strong experience with:
JWT validation
JWKS
OAuth client credentials
Secrets management
Experience integrating real-time APIs within strict latency requirements.
Understanding of fail-open vs fail-closed architecture decisions.
Ability to quickly understand and work with unfamiliar or lightly maintained open-source codebases.
Strong ability to work independently from written acceptance criteria.
Strong debugging, testing, and code-review skills.
Nice-to-Have Skills
Experience with Claude Managed Agents, including:
Agents
Environments
Sessions
Vaults
always_ask behavior
Experience authoring Claude Code plugins or skills.
Experience routing Claude workloads through:
AWS Bedrock
Google Vertex AI
Azure AI Foundry
E-commerce domain experience, including:
Product variants
Shopping cart
Checkout
Checkout hand-off
Node.js 22 experience for example storefront applications.
Experience with fraud, identity, risk, or trust APIs.
Previous Forward Deployed Engineer or customer-facing engineering experience.
Key Deliverables The successful candidate will own:
agent-trust-commerce Python package
Claude Managed Agents gateway integration
Experian MCP server
Commerce-builder companion skill
Merchant integration guide
Test and verification coverage
Retail vertical demo
Adoption-effort measurement and acceptance metrics
Candidates with general GenAI experience but without hands-on Python, Claude, MCP, and backend security experience may not be suitable for this role.