AI Developer / Builder (Lovable Expert & Claude Code) - Embedded Role

A quick bit about us, since this kind of role can be hard to picture from a job ad. Ryflow is one of Newcastle's biggest and fastest growing AI consultancies. We are based in Newcastle, New South Wales, and we work with a lot of local businesses here, mostly small and medium sized firms across professional services like law, building, engineering, project management and real estate. We build the actual AI tools and internal systems their teams use every day to run their business. We are only about a year into the business, but our client base is growing quickly and we are already the biggest AI firm in our region. As we grow, we are looking for someone to genuinely join the team, not just take on a one off project. You would be part of our culture, join our Monday meetings, and fill a real seat at the company. We want someone who will grow with us as the business grows. This is a day to day role in how we actually operate. We are looking for someone who wants to grow in three ways: their own skills, the business, and their income. As more work comes in, we will keep increasing your income. The better we do together, the better you do. All of our current client tools are built on the Lovable platform. We are also actively moving toward deeper use of Claude Code for setting up agents and more advanced workflows. This role is central to both keeping our Lovable work running at a high standard right now and helping us build real capability in Claude Code going forward. Most of our work starts with a client meeting. You turn the brief into a working tool. Right now that means working inside Lovable, scoping, prompting effectively, implementing properly, fixing issues, testing, and delivering something the client can actually rely on. Over time you will also help us build more complex agentic systems using Claude Code. You would own the build side day to day. The work includes maintaining and extending existing Lovable projects, building new tools, workspaces, agents and automations, and helping us get better and faster at the whole process. We currently do all client implementation in Lovable. Prompting correctly inside Lovable, knowing how to structure projects well, fixing bugs and errors effectively, and iterating based on real client use is essential right now. We also use Claude with custom skill files and continuously updated knowledge bases (our evolving Co-work setup) that learn from what works and what does not when prompting and implementing in Lovable. These get better over time as we feed in our judgements and learnings. We are looking to get significantly stronger at Claude Code for agents and complex automations soon, and we want someone who can help us make that transition properly. The core principle we work by is simple: AI is a powerful collaborator, but human judgement and taste are what make the real difference. We prompt well, review outputs critically, make the important decisions ourselves, and continuously teach the AI our preferences and learnings so it gets better at supporting us. We want someone who thinks and works the same way. WHAT YOU'D BE DOING - Maintain, extend and improve existing client tools built in Lovable (bug fixing, feature additions, better prompting and implementation) - Build new AI workspaces, internal tools, agents and automations, starting in Lovable now and increasingly using Claude Code as we develop that capability - Take client briefs from meetings and deliver complete, reliable tools with strong prompting and careful implementation - Get across each client's real business processes so the tool actually fits how they work - Test thoroughly and apply your own judgement so only solid, production-ready work goes to clients - Help evolve our prompting systems and skill files in Claude so the AI continuously gets better at Lovable work and agentic setups - Take initiative on tasks and drive them forward without needing to be chased WHAT WE'RE LOOKING FOR - Very strong, practical expertise in Lovable. You have built and worked on multiple real Lovable projects, know how to prompt effectively inside it, understand good project structure, and can fix errors and bugs properly. You implement things thoughtfully rather than just accepting whatever the AI outputs. - Excellent prompting skills and the ability to work collaboratively with AI tools. You bring your own judgement and taste. You use AI as a powerful tool but you direct it, review it critically, and do not outsource the important decisions and taste to the model. - Experience (or very quick learning ability and genuine interest) with Claude Code and setting up agents. We are moving in this direction and want someone who can help us build that capability properly. - You have (or are excited about) experience feeding your own judgements and learnings back into AI systems, for example through custom skill files, knowledge bases or continuous learning setups in Claude, so the AI gets better over time at supporting our specific work. - Careful and production-focused. You test properly, think about real business use cases and edge cases, and only ship work you can stand behind. - Strong written and spoken English, essential for understanding nuanced business requirements and communicating clearly with clients and the team. - High initiative and ownership. You can take a brief, clarify what is needed, build it end to end, and make good calls without heavy supervision. - Looking for a long-term role with real responsibility and the chance to grow with the company as it grows. NICE TO HAVE - A portfolio or examples of Lovable projects you have built or maintained - Experience using Claude Projects, custom instructions or skill files that evolve with learnings and feedback - Previous work with professional services clients (legal, construction, engineering, real estate, project management, finance) - Examples of how you have improved prompting approaches or AI collaboration processes over time HOW WE WORK - Human judgement and taste direct the AI. We never outsource the important decisions. - We continuously feed learnings back into our AI systems so they get better - We build things that clients can actually rely on - We keep it practical and focused on real business outcomes - We take ownership and drive work forward HOW TO APPLY We do not want long generic cover letters or AI-generated fluff answers that do not actually say anything useful. If your application reads like generic AI output with no real substance or personal judgement, we will not progress it. Instead, please send us one solid example of a Lovable project you have built or worked on significantly. For that example, tell us: - What the project does and who it is for - How you prompted and worked with Lovable (feel free to share one or two examples of prompts or approaches that worked well for you) - What you have found to be the best ways to prompt Lovable and implement projects correctly - Any bugs, errors or issues you had to fix and how you approached them - What other AI tools you currently use to help with Lovable work (for example, do you use Claude for co-work, skill files, or continuous learning setups that improve over time?) - How you have fed your own judgements, preferences or learnings back into an AI system so it got better at supporting the work (if you have an example of this) A link to the project, screenshots, repo, or a short walkthrough is very helpful. We'll be looking closely at the substance and clarity of your example, how well you explain your prompting and thinking process, your past work and reviews, and whether you demonstrate real judgement, taste and practical experience rather than generic AI output. This is the kind of detail that helps us see if you are the right fit for how we actually work.