AI Prompt Engineer — Payments AI Operating System (Fintech / ISO)

Data One Merchant Services is a payments ISO building one of the most advanced internal AI operating systems in the payments industry. We process significant annual volume and serve hundreds of active merchants — and we've spent the last several months building the infrastructure to run the entire company through a single intelligent platform. The system is built. The backend is live. The factories, boards, command centers, and agent layers are running. Now we need someone to make the AI layer exceptional. We are looking for a senior AI Prompt Engineer who can own the intelligence layer of our platform — Hermes (our AI agent), CoPilotOS (our internal assistant), and the full set of background intelligence jobs that drive our executive command center. If this goes well, we will also ask you to refer a QA Engineer you have worked with and trust — someone who can own testing and release validation alongside you. --- ## What We Built Our platform — called **Pulse / FOSM** — is a Next.js + FastAPI system with: - **Engineering Factory** — intake, risk detection, approval gates, build pipeline - **Marketing Factory** — campaign intake, compliance flagging, Chris approval triggers - **Kanban Boards** — Engineering and Marketing work items with full audit logs - **Mission Control** — executive triage hub showing KPIs, escalations, and approvals - **Jarvis Command Center** — approval queue, build queue, decision routing - **Hermes Packet Inbox** — AI-surfaced recommendations, escalations, and observations - **Javier Command Center** — activation queues, merchant follow-up, Google Sheets integration - **12 background intelligence jobs** — Hermes queue refresh, executive brief generation, deterioration detection, revenue opportunities, and more The system has a hard authority framework (A/B/C/D levels), compliance stop gates, and a rule that nothing executes autonomously without human approval. CoPilotOS is permanently scoped to Observe + Suggest in V1. --- ## What You Will Own ### Prompt Engineering - Audit all existing prompts across 12 background runtime jobs, Hermes agent, CoPilotOS assistant panel, and executive brief generator - Improve output quality, reduce hallucination risk, and enforce correct output formatting - Write and maintain system prompts, user prompts, and few-shot examples for each AI surface - Ensure every prompt that touches merchant data has explicit PII guardrails (we operate under GLBA / FTC Safeguards Rule — this is non-negotiable) - Design evaluation frameworks (evals) to measure prompt quality over time - Document every prompt with version history, intent, and expected output format ### Agent Architecture - Understand the Hermes agent layer and its role in surfacing recommendations vs. escalations vs. autonomous observations - Improve the quality and relevance of Hermes outputs - Ensure CoPilotOS assistant outputs are consistently labeled "prepared, not executed" - Design the prompt architecture for new intelligence jobs as the system grows ### QA Referral - When the engagement is established, refer a QA Engineer you have worked with and trust - Help onboard them with context on the system so they can own testing and release validation --- ## The Stack | Layer | Technology | |-------|-----------| | Frontend | Next.js 15, TypeScript, Tailwind, shadcn/ui | | Backend | FastAPI, Python 3.11, SQLAlchemy async, Alembic | | Database | PostgreSQL + Redis | | AI | Anthropic Claude API (claude-sonnet-4-6) | | Infrastructure | Docker Compose, GitHub Actions CI | | Integrations | GHL, ClickUp, QuickBooks, Google Sheets, NMI/Merchant Central | | Dev tools | Claude Code, Cursor | You do not need to be a full-stack engineer. But you must be comfortable reading Python and TypeScript code, understanding API structures, and working directly in the codebase alongside a dev environment. --- ## Who You Are **Required:** - 2+ years building production AI prompts for real applications — not demos, not chatbots, real workflows - Deep experience with Anthropic Claude API (system prompts, user prompts, tool use, structured outputs) - Understanding of AI agent architecture — the difference between retrieval, reasoning, and execution layers - Ability to write evals and measure prompt quality objectively - Strong understanding of compliance constraints in regulated industries — you know why PII guardrails matter - Experience working in a fast-moving startup with a small team — you figure things out, you don't wait for a spec - Can read and navigate a Python/TypeScript codebase without being a full-stack engineer **Strong plus:** - Experience in fintech, payments, or a regulated industry - Familiarity with background job schedulers and async AI pipelines - Previous work building internal AI operating systems or command centers - A trusted QA engineer in your network you can refer **Not a fit if:** - You only build chatbots or customer-facing conversational AI - You need a product manager to write every spec before you can work - You have not shipped prompts in a production compliance-sensitive environment --- ## Engagement Details - **Type:** Ongoing contract, part-time to start - **Location:** Remote — async friendly, overlap with US Eastern or Colombian time preferred - **Communication:** English required, Spanish a plus - **Start:** Immediate --- ## How to Apply Please include the following in your proposal: 1. **One example** of a production AI prompt you built — what it does, what problem it solved, how you measured quality 2. **Your experience with Anthropic Claude API specifically** — not OpenAI, not generic LLM work — Claude 3. **One example of a compliance or safety constraint** you built into a prompt and why it mattered 4. **Do you have a QA engineer** you have worked with and would refer? Brief description of their background. 5. **Your honest assessment** of what would make an AI agent layer excellent vs. mediocre in an internal business operating system Proposals without these five items will not be reviewed. --- *Data One Merchant Services — Medellín, Colombia / Palm Beach Gardens, FL* *Building the payments infrastructure and AI operating layer for the next generation of merchant services.*