AI Engineer
Muvo turns the texts of museums, schools and heritage sites into audio guides in up to twelve languages. You will own the AI layer: speech generation paragraph by paragraph, machine translation that respects a curator's voice, and the quality checks that make both trustworthy for a museum.
What you'll do
- Improve the speech pipeline: prompts and voice direction per language, pronunciation of names, dates and acronyms, loudness and silence handling with ffmpeg.
- Evolve AI translation: glossaries per space, consistency across paragraphs, and evaluation sets that catch regressions before customers do.
- Design the background jobs that generate, retry and publish audio with idempotent credit accounting.
- Measure quality and cost per paragraph and turn the numbers into product decisions.
- Work directly with museums and teachers to understand what a good audio guide sounds like.
What we're looking for
- Professional experience shipping LLM or speech features to production, not only prototypes.
- Strong TypeScript and Node.js; comfortable with SQL and PostgreSQL.
- You write evaluations and tests, and you care about latency, cost and failure modes.
- Fluent Spanish and English, written and spoken.
Nice to have
- Audio processing (ffmpeg, loudness normalisation, codecs).
- Experience with Next.js App Router and React Server Components.
- Knowledge of Catalan, Basque, Galician, French, German, Italian or Japanese.
- Interest in museums, education or cultural heritage.
What we work with
- TypeScript
- Next.js
- PostgreSQL
- Drizzle ORM
- Supabase
- Inngest
- LLM APIs
- Text-to-speech
- ffmpeg
- Vercel