AI & Automation Specialist (Claude, Prompt Engineering, n8n, Microsoft 365)

About Us:
We're a 250-employee US multi-brand retail and wellness company operating 22+ locations. We've been in business more than 22 years and we grow by building strong teams and better systems — not by chasing trends.

We're not a startup, and we're not an agency. We're an established operating company that has d ---------- AI is a real operational advantage, and we've already proven it on a small scale. Now we want to do it properly across the business.

The Opportunity
We've built working AI prototypes that help our district managers review daily operational reports, check policy compliance, and track team performance. They work. Our managers use them. But they were built by operators, not engineers — and they don't scale.

Your job is to turn those prototypes into reliable, company-wide systems, and then keep finding new places in the business where automation removes low-value work.

You'll be the first dedicated AI person we've hired. You will not be maintaining someone else's mature stack — you'll be building ours. If you want ownership and visible impact, this is that role. If you want a well-defined ticket queue, this isn't.

What You'll Own
Refine and scale our existing AI assistants. Our current systems are built on Claude and Chat GPT. Improve prompts, structure the underlying reference material, and make outputs consistently useful to a busy district manager reading on a phone between store visits.

Build the knowledge base. Organize and maintain the operational content our assistants draw from — policies, procedures, troubleshooting guides — inside our Microsoft 365 environment, so answers stay current when the underlying policy changes.

Automate recurring workflows. Find the repetitive administrative work happening across operations, marketing, and accounting, and eliminate it with automation. We'll point you at the first several; after that we expect you to find them yourself.

Generate documents automatically. Build systems that turn AI output into clean, print-ready PDFs and Word files that match our formatting standards — so a manager gets a finished document, not text to reformat.

Improve accuracy and reliability. Design a testing process that catches wrong or outdated answers before our managers see them. Track error rates. Reduce them. Tell us honestly what the system can and can't be trusted with.

Evaluate what's next. Claude is our Sr. Leadership's primary platform, and we intend to keep building on it. Your job here is to stay current on what it can do, and to tell us when a supporting tool or model is worth adding for a specific job.

Document everything. Every system you build needs written documentation good enough that someone else could maintain it. This is a hard requirement, not a nice-to-have.

Requirements
2+ years building AI assistants, custom GPTs, or LLM-powered tools that real people actually used in production. Personal experiments and course projects don't count. We will ask you to walk us through something you shipped.
Hands-on prompt engineering experience, including structuring reference material so a model answers from source content rather than guessing.
Working knowledge of Microsoft 365 — SharePoint and OneDrive file structure, and how to get content out of them programmatically.
Practical automation experience with n8n, Make, Zapier, or Power Automate. You've built multi-step workflows that run unattended without breaking.
Python at a level sufficient to write and debug scripts for data cleanup, API calls, and document generation.
Hands-on, current experience with Anthropic Claude. Claude is our primary AI platform and the environment you will work in every day. You need to be genuinely comfortable in it — Projects, custom instructions, artifacts, and how Claude behaves differently from other models. Familiarity with additional platforms such as ChatGPT or Google Gemini is useful, but Claude proficiency is not optional here.
Strong written English. You'll write documentation, prompts, and process guides that US managers read without translation help.
Reliable setup: dedicated work computer, 25+ Mbps primary internet, and a backup connection.

Nice to Have
Microsoft Graph API experience
Retrieval / RAG implementation experience
Document templating (docx or PDF generation from templates)
Experience supporting non-technical field managers, not just office staff
Retail, restaurant, franchise, or multi-location operations background

What We're Not Asking For
We're not looking for someone to promise us a system that's never wrong. Language models make mistakes. What we want is someone who is honest about where the risk is, builds checks around it, and tells us when something isn't ready. Candidates who claim perfect accuracy will not advance.

Schedule
Core hours: 8:00 AM – 12:00 PM US Central Time, Monday through Friday — required.

In Manila, that is approximately:

9:00 PM – 1:00 AM (mid-March through early November, US daylight saving)
10:00 PM – 2:00 AM (early November through mid-March)

The remaining hours of your workday are flexible — work them when you focus best. We deliberately kept the required overlap to four hours because deep build work goes better uninterrupted, and because we'd rather you have a sustainable schedule than burn out on a full graveyard shift.

Please confirm in your application that this schedule works for you long-term.

How to Apply
Read this section carefully. Applications that skip any step will not be reviewed. We'd rather receive 10 thoughtful applications than 200 copy-pasted ones.

1. Start your application with the code word: LEVERAGE

2. Answer these five questions in writing. Be specific and brief — a few sentences each is fine.

a. Describe one AI assistant or automation you built that real people used. What did it do, what did you build it with, and what broke that you had to fix?

b. A manager asks our AI assistant about a company policy. The assistant answers confidently using a version of the policy that was updated six months ago. How would you prevent that from happening again?

c. Which automation tool do you reach for first, and what's a situation where you would not use it?

d. Tell us specifically how you've used Claude — which features, for what, and how recently. Then tell us one thing Claude does differently from other models that actually changed how you built something.

e. What is Claude not good for, in your experience? Where would you use something else, or nothing at all?

3. Record a 3-minute Loom video (or similar) walking through something you built. Show the actual work — the workflow, the prompt structure, the output. Explain why you built it that way. If any of it was built with Claude, show us that. Screen share is more useful to us than seeing your face.

4. Tell us your monthly rate expectation and your earliest start date.