About

💡 Seeking Specialist: Audio ML Engineer for Passion-Driven POC Help me (a solo entrepreneur), build the core engine for an innovative consumer audio project. About the Project & The Vision: I am a solo entrepreneur with a deep passion for music; trying to launch a solution that will hopefully have big impact for music enthusiasts; and I need a technical expert to validate the critical core - the Proof of Concept (POC). Your work will be essential in moving this passion project from a concept to a viable product. I'm looking for a partner, not just a contractor, for this initial, crucial technical challenge. 🎯 The Core Technical Challenge The project requires solving a complex, niche problem in signal processing and machine learning. We need an ML pipeline that can: - Input Data: Analyze noisy, real-world audio files containing multiple overlapping signals (i.e., a primary structural component mixed with background noise and a distinct secondary source). - The Problem: The model must accurately isolate and transcribe the specific structural component (our target signal) while effectively neutralizing the high noise levels and interference from the secondary source. - Goal: Achieve a benchmark accuracy of 85%+ for classifying and time-aligning the features of the target structural signal. - The Deliverable: A robust, low-latency API endpoint (using Flask/FastAPI) that serves the POC model. 📝 Required Expertise & Skills I am looking for an engineer with highly specialized skills in: - Deep Domain Expertise: Proven experience in Audio Machine Learning and Digital Signal Processing (DSP). - Programming: Expert proficiency in Python (NumPy, SciPy) and ML frameworks (PyTorch/TensorFlow). - Audio Tools: Direct experience with specialized audio libraries used for Music Information Retrieval (MIR) or complex signal separation. - Deployment: Practical experience serving ML models via APIs (Flask/FastAPI) for low-latency inference. 🤝 How to Apply To apply, please submit your proposal. I ask that you include a brief, technical description of a past project where you successfully separated a target structural signal from a multi-source, noisy audio input and detail the specific libraries/methodologies you used. Confidentiality Note: Due to the proprietary nature of this passion idea, I will require a signed NDA before sharing the full technical specification and detailed data sets. I look forward to hearing your insights!