Full Stack Engineer (mfd)

You can take a problem further than the ticket. Work often starts with something incomplete like customers need a better way to manage X. Youll shape the problem with product challenge assumptions and decide what belongs in this iteration. Strong opinions about what we build not onlyhow are expected here.

You care about product. You want to know who youre building for why the problem matters and what success looks like. That context should shape your technical decisions.

You have technical judgement. There is rarely one correct architecture. You know when the boring solution beats the ambitious one when an abstraction helps and when it hides the problem and when a small feature is about to create six months of technical debt. We expect you to reason about API boundaries data models permissions performance failure modes testing observability and rollout and to raise concerns with a clear argument attached.

You make the system better not just the feature. You fix the spreading pattern rather than the single instance make an API easier for the next person and improve tests because the current setup makes everyone afraid to deploy. Your impact should show beyond the pull requests with your name on them.

You can work across the stack. A feature might run from a React interaction through GraphQL into a Node service a schema redesign a permissions update and a safe migration for existing customers. Curiosity to follow the problem matters more than equal depth everywhere.

You own quality. We have QA. Thats not where quality begins. What breaks if this request happens twice What happens with old data or when the API fails halfway through We use Playwright and Jest alongside reviews and monitoring. The goal isnt coverage its confidence.

You communicate like your decisions will matter later. Six months from now somebody should understand an important decision without digging through seventeen Slack messages and a deleted Notion comment. That means writing things down when they matter explaining trade-offs clearly giving useful code reviews and surfacing risks before they become incidents.

You raise the engineering bar around you. You do not need a management title to have influence. Find the hole in an architecture unblock someone without taking the keyboard away disagree without turning every decision into a referendum. As you earn trust youll naturally become the person people bring harder problems to.

• Substantial production experience with TypeScript React and
• Comfortable reasoning about relational databases not just the ORM
• Can independently take a feature from unclear starting point to production
• Can read unfamiliar code without immediately proposing a rewrite

Nice to have:

• Experience with GraphQL Prisma or PostgreSQL at scale
• Exposure to LLM APIs embeddings or vector search
• Experience owning systems not just features

Dont check every box Apply anyway. Wed rather have the conversation than lose someone over a bullet point.

Were a small engineering team so there are fewer places for responsibility to disappear and no chain of five people between you and a decision. We prefer written context over meetings the calendar invented and working software over elaborate process.
Quickly doesnt mean carelessly. Sometimes the fastest route is the simple version today sometimes its another day on a foundation twenty future features will sit on. Knowing which situation youre in is part of the job. Expect to ship to production very early.
No seven-stage endurance test. No algorithm trivia disconnected from the work.

1. Intro. You us and whether this is the environment youre actually looking for.
2. Technical conversation. We go deep on something difficult youve built: the decisions the alternatives what youd do differently now.
3. Technical assessment. Usually live coding on a realistic problem sometimes a small proof of concept instead. We care how you think not whether your solution matches ours.
4. Offer. If theres strong mutual conviction we move.

One last thingJob descriptions have a habit of producing candidates who optimise themselves against bullet points. Please dont. If youre reading this thinking I can do most of this but I havent worked with MCP apply.
Strong fundamentals meaningful software shipped and a bias toward ownership are what were looking for. The rest is learnable.