I focus on taking complex ideas from concept to working software. I lean heavily into clean engineering, deliberate system design, and building structured, maintainable machine learning models.

Lyra is a music audio-to-sheet-music algorithm designed to process audio signals and automatically transcribe them into digital sheet music notation.

1. Audio Ingestion & Processing (NumPy): Loading raw audio waveforms and computing Short-Time Fourier Transforms (STFT) to transform time-domain audio data into frequency spectrogram arrays.
2. Deep Learning Inference (PyTorch): Passing processed feature matrices through an optimized neural network architecture to predict musical pitches, note durations, and onset timings.
3. Visualization & Metrics (Matplotlib): Plotting model training loss, validation metrics, and alignment matrices for system debugging.
4. Data Infrastructure (Supabase): Managing storage pipelines for processing queues and persistent data storage.

• Collaboration: Need a hand reviewing, optimization-testing, or debugging Python code? Shoot me a message!