Backend-Focused Engineer for AI Audio Platform

We are building a new AI-powered audio stem separation and enhancement platform (complementary to tools like Suno). We need a strong backend-focused engineer to develop the core audio processing pipeline, including: Audio file upload handling High-quality stem separation (vocals, drums, bass, other, etc.) Advanced stem enhancement: artifact reduction, spectral repair, harmonic reconstruction, bandwidth extension, and techniques to improve sparse harmonics / thin audio common in AI-generated tracks Clean export/download of processed stems Scalable architecture (async processing, GPU support, queueing) Requirements: Strong Python experience Hands-on work with audio ML models (HT-Demucs or similar, torchaudio, Pedalboard, etc.) Experience implementing audio post-processing and enhancement techniques FastAPI or similar for backend API Ability to deliver clean, well-documented, production-ready code quickly Nice-to-Haves: Experience with Celery / RQ for async workloads GPU inference optimization Familiarity with music generation tools (Suno, Udio, etc.) Scalable cloud infrastructure (AWS or Azure) This is the foundational "moat" component of the platform. We are looking for someone who can own the backend audio processing end-to-end and potentially grow into a longer-term role.