About
Hi, I'm Pierre
I'm a computer science engineer who builds and ships products. I spend most of my time writing code, talking to users, and figuring out how to make things actually work in production.
I'm an auditive learner — I retain information much better when I hear it than when I read it. Podcasts, lectures, audio books: that's how I consume content. But most of the things worth reading don't come in audio form. So I built Podcastify to fix that for myself — and apparently for a lot of other people too.
What Podcastify does
Podcastify turns any piece of content into a produced two-host podcast episode in under three minutes. Paste a URL, upload a PDF, drop in an image, or write raw text — the platform extracts the content, generates a natural back-and-forth conversation with an LLM, then voices it with neural text-to-speech.
- Articles, blog posts, newsletters, and websites
- Research papers and whitepapers (PDF)
- Meeting notes and presentations
- Raw text or ideas you want to think through out loud
The stack
The pipeline runs on FastAPI + Celery workers, uses Google Gemini for transcript generation via LangChain, and supports multiple TTS providers — Gemini, OpenAI, ElevenLabs, and Edge. The frontend is Next.js with Supabase for auth and storage. Everything is containerised and deployed on a single docker-compose stack.