# Confidential — AI Agent Architect

- Generated: 2026-08-25 10:58:30 AM EDT
- Original/canonical posting: https://www.linkedin.com/jobs/view/4458509897/
- Posted: 16 hours ago at capture time
- Applicants: Be among the first 25 applicants
- Work model/location: United States; exact remote/hybrid designation not disclosed
- Employment type: Full-time
- Compensation: Competitive base and bonus; no numeric range disclosed, so Architect title does not qualify
- Travel: Not disclosed
- Positioning track: Technical IC
- Fit outcome: FAIL — 79%

## Direct-match strengths

Codex-driven production development, agent orchestration, prompt/governing documents, LangChain/LangGraph, RAG, full-stack architecture, standards, and agent-output review are unusually strong.

## Hard or material gaps

Hard compensation/work-model gates: no numeric pay range is disclosed for an Architect title, and remote or Massachusetts eligibility is not established. The JD emphasizes Claude Code, Copilot Workspace, Cursor, and Devin across React, gRPC, and Terraform; Keith has strong Codex/Claude agentic coding evidence but not the complete named tool/stack breadth.

## Weighted evidence map

1. AI coding agents shipping production software — 3/3: More than half of AssistX was built with code-generation technologies including Codex.
2. Agent prompt and governance documents — 3/3: Direct skill/system-prompt, guardrail, and controlled-prompt experience.
3. Agent frameworks and RAG — 3/3: Direct LangChain, LangGraph, agents, and RAG.
4. Full-stack technical judgment — 3/3: Direct Python, APIs, databases, AWS, UI, and platform delivery.
5. Claude/Codex orchestration — 3/3: Direct supported evidence.
6. Devin/Cursor/Copilot Workspace and full named stack — 1/3: Not fully supported.
7. Compensation and eligible work model — 0/3: Both fail or remain unverified.

## Keyword diagnostic

No resume generated. The functional fit is strong, but compensation, work model, and named-tool breadth prevent PASS.

## Full normalized job description

AI Agent Architect (AI-Native Builder & Agent Governance Lead)
Market-leading property management firm seeks an AI Agent Architect to deploy AI agents directly within and around the SaaS platform running the firm today.
You'll own two things end to end:
🔹 Agent Orchestration — Direct tools like Claude Code, GitHub, Copilot Workspace, Cursor, and Devin to build and extend features across the platform's full stack (database, APIs, gRPC, React, Terraform). Decompose business requirements into agent instructions. Review and validate agent output with real technical judgment.
🔹 Agent Governance — Author and continuously refine the markdown-based system prompts, guardrails, and behavioral documents that define how every agent reasons and what it's permitted to do. Treat these documents with the same rigor as production code.
What we're looking for:
✅ Demonstrated experience directing AI coding agents to ship production software — not primarily writing code by hand
✅ Expert-level prompt engineering and governing-document authorship
✅ Experience with an agent orchestration layer spanning multiple technology layers
✅ Hands-on experience with a major agentic framework (LangChain, LangGraph, CrewAI, AutoGen, Semantic Kernel, or similar)
✅ Strong grasp of RAG architecture
✅ Strong software engineering foundations — the judgment to know when agent output is right, and when it isn't
✅ Comfortable setting technical direction and standards without formal management authority
Why this role: Real ownership over how AI is governed at scale, on a proprietary platform with genuine operational complexity — not a demo environment.
Competitive base, bonus, and full benefits.

## Artifact metadata

- Resume and cover letter: Not generated under the fit gate.
- LinkedIn connection note: Not generated for a FAIL role.
- Google Drive used: No