Behind every answer MagicInterview shows is a pipeline: audio becomes text quickly, a model reads the conversation and your sources, and a short, speakable answer comes back before the pause gets awkward. You will build and own that pipeline and the product around it.
What you will do#
- Work across the Node.js server, the React web app and the data behind them, including accounts, sync, projects and sources.
- Keep live coaching fast and dependable: speech recognition, speaker detection, model routing, retries, timeouts and the checks that keep answers short and truthful.
- Build features such as Research Online, practice with AI voices and the connection for AI assistants, from the model call to the screen.
- Decide how we use cloud models and the device's own models, and measure what each does for quality, latency and cost.
- Look after production: observability, error handling, safe releases and the database.
- Keep what we say about privacy true, in code. Audio is not stored, and data goes only where a feature needs it.
What we are looking for#
- Strong JavaScript or TypeScript on Node.js and React, and experience running a production service.
- Experience with LLM products: prompts that behave, structured outputs, evaluating answers, and dealing with provider errors and rate limits.
- Real-time instincts: streaming, latency budgets, backpressure and failure modes.
- SQL fluency. We use MySQL.
- A habit of testing what you ship and checking the live result.
- Clear writing.
Nice to have#
- Speech systems: Whisper-class models, voice activity detection or speaker embeddings.
- OAuth and the Model Context Protocol.
- Experience with Expo or native mobile APIs.
- Multilingual products and testing across languages.
How we work#
Remote-first, Europe time zones, flexible hours. Small team, written decisions, small releases. The whole stack is open to you.
How to apply#
Email support@magicinterview.app with "Application: Full-stack Engineer" in the subject. Send a link to something you built, ideally something with AI or real-time pieces, and a few lines about why this role. Other roles are on the Careers page.