AI Product Builder

Kolomolo

1d•Remote

About The Position

We're looking for someone who can take a business problem such as :"we need an ERP", "we need a project tracker", "we need a client portal" and build a working application end-to-end using AI development tools, primarily Claude Code. This isn't a traditional developer role. You won't be writing code line by line. You'll be directing AI to build, test, and ship real software fast. Think of it as a solution architect who actually delivers the solution, using AI as their engineering team. There is no established job title for this yet. The closest analogies are a technically-grounded product manager, a hands-on solution architect, or a full-stack developer who has made the leap to AI-native building. What matters is the output: working software, built quickly, that solves real business problems.

Requirements

Demonstrated AI building experience
You have applications you've built using AI coding tools that you can show us. Not toy projects functional, usable software that solves a real problem.
You understand the workflow: how to prompt effectively, how to structure a project so AI can work with it, how to catch and correct when the AI goes wrong.
Applications must be demo able. Screen recordings, live demos, or deployed links. If you can't show it, it doesn't count.
You don't need to be an expert programmer, but you need solid working knowledge across:
Languages: What Python is good for vs. TypeScript vs. Go. When a compiled language matters. You can read code confidently even if you don't write it from scratch.
Frontend frameworks: React, Next.js, Vue, or similar. You know what a component is, how state management works, what makes a UI performant.
Backend frameworks: Express, Fast API, Django, or similar. You understand REST APIs, authentication flows, middleware, and how a request moves through a server.
Databases: When to use PostgreSQL vs. a document store vs. SQLite. You understand schemas, indexing, migrations, and can design a sensible data model.
Testing: Unit tests, integration tests, end-to-end tests. You know what each catches and when each matters. You've used testing to catch regressions in AI-generated code.
DevOps basics: You can deploy an application. Docker, basic CI/CD, cloud hosting (AWS, Vercel, Railway, or similar). You don't need to be a sysadmin, but you can get software running in production.
You can take a vague business need and decompose it into buildable pieces.
You make pragmatic scope decisions you know when 80% is enough and when the last 20% matters.
You think about the end user. The software you build is usable, not just functional.
You can communicate clearly with non-technical stakeholders about what's possible, what's hard, and what trade-offs exist.

Nice To Haves

You've built something ambitious on your own an ERP module, a CRM, a project management tool, an internal business system using AI tools.
You learn new domains fast. You can go from "I've never thought about warehouse logistics" to "here's a working inventory management system" in days, not months.
You have opinions about AI-assisted development: what works, what doesn't, where the limits are, and how the workflow is evolving.
You've hit the walls of AI coding tools and found ways around them context limits, hallucinated APIs, architectural drift. You have strategies, not just frustrations.

Responsibilities

Take loosely defined business requirements and turn them into working applications independently, from first conversation to deployed product.
Use Claude Code (and similar AI coding tools) as your primary development environment. You direct the architecture, the AI writes the code, you validate and ship.
Scope and prioritise features. Decide what to build first, what's good enough for now, and what needs to be robust from day one.
Make sound architectural decisions: choosing the right database, the right framework, the right deployment approach for each project.
Test what you build. Understand what kinds of testing matter for each context and when you need unit tests, when integration tests matter more, when a manual smoke test is enough.
Work across a range of business domains. One week it might be a logistics tool, the next an internal reporting dashboard, the next a customer-facing SaaS product.