Full Stack Developer – AI Agents & Workflow Systems

About Lakestone.ai

Lakestone.ai is Perth's fastest growing AI consulting firm building AI-powered systems for our clients.

We dont sell SaaS products. We build bespoke workflow systems, dashboards, portals, automations, agents and AI-assisted tools.

Our work combines full-stack development, workflow analysis, AI tooling, database design, integrations and strong product judgement.

About the Role

This is a hands-on role for a curious problem solver who can work with clients to uncover useful AI opportunities, build and ship real agents and workflow automations, and take a solution from an unclear starting point through to something running in production. You'll also build the full-stack applications (dashboards, portals, forms) that these agents live inside.

Learning speed, judgement and communication matter as much as deep engineering specialisation. We use Claude Code as our primary tool and expect strong fluency with it — but what we're really hiring for is broader: real experience designing and shipping agents, in whatever framework you used to build them.

This role is ideal for someone who can:

Ask the right questions and turn an unclear business problem into a useful, buildable AI opportunity.

Build, test and ship practical agents and automations.

Explain their thinking, trade-offs and limitations clearly, to technical and non-technical people alike.

Take ownership from discovery through to delivery and a measurable outcome.

Key Responsibilities

AI Opportunity Discovery

Work with clients to explore business problems and identify practical, valuable AI use cases.

Ask targeted questions to understand existing workflows, constraints and data.

Turn a messy starting point into a defined, testable problem connected to real value.

Agent Prototyping & Delivery

Build, test and improve agents and automations — document extraction, RAG, tool-calling, multi-agent workflows — using modern agent frameworks and Claude Code.

Make sensible calls on model selection, prompt design, cost/token usage and memory/state persistence as you build.

Troubleshoot issues, test assumptions and refine an approach until it produces a reliable, practical outcome.

Full-Stack Delivery

Build the front-end and back-end that sit around each agent — dashboards, forms, APIs and the data behind them — using modern TypeScript / JavaScript.

Ship into real production systems, with the judgement to know when to slow down and get it right.

Client Communication & Adoption

Communicate progress, decisions and constraints clearly, and set realistic expectations about what an agent can and can't do.

Take work from prototype through to something the client actually adopts and trusts.

Requirements

Proven experience building and shipping real full-stack applications — not just isolated components.

Practical experience building and deploying AI agents or LLM-enabled automations — personal, employer or client work, any framework.

Strong commercial TypeScript/JavaScript; Python a plus.

Strong hands-on experience with Claude Code — you'll be tested on your setup, workflow and skills.

Confident communication — able to explain technical decisions and trade-offs simply to both technical and non-technical people.

Strong curiosity, structured problem-solving, and the ability to learn new tools fast.

A portfolio, GitHub profile or case studies that show what you've actually built and shipped.

Nice to Have

Experience with document extraction, RAG, or agent frameworks/SDKs (Claude Agent SDK, LangChain/LangGraph, OpenAI Agents SDK, MCP).

Bonus points for Supabase and Trigger.dev.

Consulting, agency or multi-client delivery experience.

Experience working directly with business owners or operations leads.

A quality-first mindset, including testing agent reliability, not just UI code.

What Success Looks Like

The real problem is clearly understood, and the AI use case is practical and testable.

Solutions move from prototype to something genuinely running in production, not stuck as a demo.

Clients understand what the solution does, its limitations, and how to use it confidently.

You've taken ownership from an unclear starting point through to a measurable outcome.

Culture and Ways of Working

Small but experienced team.

High trust, high pace, high standards — practical over theoretical, AI-native, strong engineering foundations.

We move quickly, but anything reaching production is secure, reviewed and maintainable.

Office-first while the team and pipeline are growing fast, with flexibility where it makes sense.

What We Offer

Competitive salary with bonus/incentive packages.

Real ownership of client systems and the agents inside them — not narrow tickets.

Modern full-stack and AI agent tooling as your daily toolkit.

Direct exposure to real business workflows across established Australian companies.

A small, visible team where your contribution matters.

Location

Perth-based only. Regular client-site attendance required. Candidates must have full working rights in Australia to be considered.

How to Apply

Email your resume and examples of what you've actually built — portfolio, GitHub, case studies, walkthroughs — to ••••@lakestone.ai.

Briefly cover:

Full-stack applications you've built.

Agents or workflow automations you've built and shipped.

Your experience using Claude Code or similar AI coding/agent tools.

A project where you took ownership from problem through to delivery.