AI Operations Manager

Moburst; Johannesburg; On-site; ZAR 550,000 - 850,000; Full time

Job summary

Moburst is hiring to own the operational side of our AI team's output, maintaining existing automations and building small, well-scoped automations from an existing spec. You’ll ensure reliability, keep credentials current, and document changes for easy handoff.

You will work hands-on, testing edge cases and thinking like the user to prevent breakages before users encounter them. Strong scripting with Python or JavaScript is a plus.

Qualifications

• Hands-on experience building automations in a low-code platform.
• Experience owning monitoring, error handling and documentation for a system.
• QA mindset: identify edge cases and prevent user-impacting failures.
• Empathy for real user workflows in assessment of builds.
• Scripting ability (Python or JavaScript) to extend automations.
• Organized to maintain a registry/inventory as standard practice.

Responsibilities

• Monitor and maintain automations already shipped; catch breakages.
• Build new automations from defined specs using n8n, Make, Zapier.
• Maintain live inventory of running automations, dependencies, owners.
• Triage day-to-day fix or small-build requests; escalate when design is needed.
• Document every touched automation sufficiently for handoff.
• Test like QA: cover edge cases and fix before release.
• Act as the user to flag confusing or fragile parts from client workflow.

Skills

Low-code automation

QA mindset

Python

JavaScript

User empathy

System monitoring

Documentation

Tools

n8n

Make

Zapier

API integration

Job description

Moburst is a full-service, mobile-first digital marketing agency that helps companies exceed their growth goals. Leading global brands and startups (including Google, Samsung, and Uber) leverage our mobile and digital expertise to grow their products, strengthen their market presence, and maximize performance KPIs.

We're hiring someone to own the operational side of our AI team's output: keeping what's already built running, and building the smaller, well-defined automations that don't need a full system design pass. Where our AI builders design and ship new end-to-end systems from open client use cases, this role keeps the existing estate healthy and builds from a spec that's already written.

This is a hands-on, detail-first role. You'll be the person who notices an automation is about to break before anyone else does, and who can turn a well-scoped request into a working build without needing it handed to you twice.

Responsibilities:

Monitor and maintain automations the AI team has already shipped: catch breakages, keep credentials and integrations current, and fix issues before anyone else notices them.

• Build new automations from an already-defined spec (n8n, Make, Zapier, or equivalent), for tasks that are scoped, as distinct from open-ended use case design.
• Maintain a live inventory of what's running, what it depends on, and who owns it, so renewal dates and credential locations don't rely on any one person's memory.
• Triage day-to-day fix or small-build requests from the team, escalating anything that needs full system design.
• Document every automation you touch well enough that someone else on the team could pick it up without a handoff call.
• Test like a QA engineer: before anything ships, push it through edge cases (malformed or empty inputs, expired tokens, rate limits, API changes, unexpected volume) and fix what breaks before it breaks for a user.
• Be our user: use every build the way the account managers, strategists, and clients who depend on it would, and flag anything confusing, slow, or fragile from their side of the screen.

Requirements:

• Hands-on experience building and maintaining automations in at least one low-code platform (n8n, Make, Zapier, or equivalent); n8n experience preferred.
• Comfortable owning the less visible parts of a system: monitoring, error handling, documentation, credential hygiene.
• A QA mindset: you instinctively ask “what happens if…”, hunt for edge cases, and would rather find the failure yourself than hear about it from a user.
• User empathy: you can think like our ICP (Ideal Customer Profile) and judge a build by whether it works for the person relying on it, in their real workflow.
• Enough scripting ability (Python or JavaScript) to extend an automation when a platform's native nodes run out.
• Organized enough to maintain a registry or inventory as standard practice, not only when asked.

Nice to have:

• Prior ops, support engineering, or QA background, where the job was keeping something running rather than designing it new.
• Basic familiarity with Claude tooling (API, Claude Code).