Full Stack AI Engineer (Front-end leaning)

About the role

Interview Resources turns a company name into an evidence-graded interview report. Every question we show is labelled by how well it is grounded, and every claim carries the source it came from. That honesty is the product, and most of it is a front-end problem: a model can produce a confident paragraph in a second, and the interface has to make clear which parts of it are earned.

This role owns that surface. You will build the research form, the streaming run view, and the report itself, and you will go as deep into the backend as the feature needs. Front-end leaning means where you spend most of your week, not a wall you stop at. The work is close to the model, so you will spend real time on streaming, partial states, retries, and the long tail of ways a generative flow fails in front of a paying user.

We are small. You will pick the components, name the routes, and decide what a screen looks like. Nobody is going to hand you a Figma file for every ticket.

What you will do

• Design and build product surfaces end to end in Next.js and React, from the route and the data fetch through to the last hover state.
• Make streaming and long-running AI work legible: progress that means something, partial results that stay readable, and failures that explain themselves.
• Turn a graded report into an interface a candidate can scan in two minutes and trust, including the parts where we say we do not know.
• Build the server routes, database queries, and schema changes your features need, in Drizzle and Postgres.
• Hold the line on accessibility, keyboard behaviour, dark mode, and mobile. These are requirements here, not a polish pass.
• Write tests that survive refactors: Vitest for units and components, Playwright for the flows that take someone's money.
• Watch what real users do after you ship, then fix what the data says is broken instead of what is fun to fix.

What we are looking for

• Three or more years building production web applications in TypeScript, with a large share of it in React.
• Real, current experience with a modern React framework and its server rendering model. We run Next.js App Router.
• You can build a UI from a rough description and a product goal, without a pixel-level design to trace.
• Comfort writing SQL and reasoning about a relational schema, not only calling an ORM and hoping.
• You have shipped something that talks to an LLM API in production, and you know why the second version was different from the first.
• Clear written communication. Most of what you decide here gets decided in writing.

Nice to have

• Strong visual instincts and a portfolio of interfaces you are proud of.
• Experience with streaming responses, server-sent events, or the AI SDK.
• Familiarity with Tailwind CSS v4 and modern component primitives.
• You have worked at a company small enough that you also answered support tickets.

How the process runs

• Intro call, thirty minutes, with the founder.
• A paid take-home you can finish in a focused afternoon, built on the real stack.
• A ninety minute working session on your take-home: we extend it together.
• A final conversation about scope, money, and what the first ninety days look like.