Machine Learning Engineer | Audio & Music AI

Job Description

To develop the intelligent features of AmpLab, a software product in the field of guitar and music processing, we are looking for a machine learning specialist.
The training infrastructure and initial prototypes already exist in the project. The goal is to improve model quality and turn the models into practical, usable product features.

Areas of work:
• Separating instrument and vocal sounds from a song (Stem Separation)
• Note and chord detection, and converting guitar audio to tablature and MIDI
• Training neural models to simulate the behavior of amps and effects, and the impact of their settings
• Research and development of new music analysis and understanding features
Depending on your expertise, the collaboration will begin in one or more specific areas; mastery of all of the above is not required.
Responsibilities:
• Reviewing existing models and research, and choosing a suitable approach for each problem.
• Preparing audio data, labeling, generating synthetic data, and increasing data diversity.
• Training, tuning, and comparing models, and analyzing the causes of their errors.
• Evaluating with independent data, technical metrics, and listening tests.
• Optimizing runtime speed, memory usage, and computational resources, and collaborating on integrating the models into the software.
• Logging experiments and documenting results and limitations.

Required skills:
• Proficiency in Python and PyTorch.
• Hands-on experience in training, evaluating, and improving deep learning models.
• Ability to work with datasets, detect overfitting, and design reliable evaluations.
• Ability to read and implement papers and open-source projects.
• Ability to work independently, solve problems, and give clear progress reports.

Bonus skills:
• Experience in audio, speech, or time-series signal processing.
• Background in sound separation, note and chord detection, or neural audio modeling.
• Familiarity with tools such as Demucs, RoFormer, Basic Pitch, or Neural Amp Modeler.
• Experience with GPU training, ONNX export, or model optimization for low-latency execution.
• Familiarity with guitar, music theory, tablature, or music composition software.

Specialized experience in music or playing an instrument is not required. We also welcome specialists from other areas of deep learning who have a strong portfolio and an interest in working on audio and music problems. The main criterion is practical ability to develop models and achieve measurable improvements.

Type of collaboration: Remote
Salary paid in Australian dollars