Organization Overview
Reality Labs Research at Meta is a leading research organization of world-class researchers, developers, and engineers who collaborate to actively create a future where virtual and augmented reality become as indispensable as todays smartphones and personal computers.
The mission of the Reality Labs Research Audio team is to engineer augmented audio that redefines human hearing capabilities. This will allow us to connect people by facilitating conversation in even the most challenging auditory environment.
Role Summary
We are seeking a contract applied research scientist specializing in building machine learning (ML) models that emulate auditory perception. This role is an integral part of our team, contributing to our research and development efforts. The ideal candidate will help us explore and understand individualized audio quality preferences and experiences, enabling us to tailor our technologies to meet unique user needs.
Responsibilities
• Drive research on improved machine learning models for speech quality • Run computational experiments and report findings • Implement ML models that emulate aspects of human auditory perception • Develop next-gen audio quality models • Independently implement ML training pipelines, models, and evaluation frameworks • Regularly report on project progress, dependencies, and risks to stakeholders • Support research scientists and engineers within the team • Execute on applied coding tasks in support of the teams goals
Minimum Qualifications
Interpersonal and communication skills with strong attention to detail. Proactive with ability to execute on multiple projects simultaneously. Strong organizational and time management skills. Track record of communicating research on ML perception in an academic or industrial setting. Experience working in a fast-paced research and/or product development environment. Experience with Python or other scientific programming languages. Previous experience designing training and evaluating neural networks in PyTorch. Experience working in a high-performance computing environment. Masters degree or equivalent experience in Computational Neuroscience, Cognitive Science, Electrical Engineering, Perception, Experimental Psychology, Audio Engineering, Acoustics, Computer Science, Computer Engineering, Biomedical Engineering, or a related field.
Preferred Qualifications
PhD degree or equivalent experience. Experience with modeling audio quality.
Top 3 Must-Have HARD Skills
• Deep Learning for Audio • Speech Processing
The role requires building ML models for speech quality assessment and auditory perception. The candidate must have hands-on experience designing training and evaluating neural networks specifically for audio applications using PyTorch, TensorFlow, Keras, or other desired frameworks.
• Psychoacoustic & Perceptual Modeling Expertise
The candidate must understand how humans perceive audio quality, including concepts like speech intelligibility, listening effort, speech degradation, and noise noticeability. This is critical for building models that accurately predict subjective Mean Opinion Scores (MOS) and enable features like Conversation Focus to be evaluated computationally rather than through time-consuming user studies.
• ML Training Pipeline & Evaluation Framework Development
The role requires independently implementing end-to-end ML pipelines, data preparation, model training, hyperparameter tuning, and evaluation using metrics like Mean Absolute Error (MAE), Pearson correlation, Signal-to-Noise Ratio (SI-DR), PESQ, and STOI. Experience with High-Performance Computing (HPC) environments for large-scale training is essential.
Good-to-Have Skills
• Binaural audio processing and spatial audio quality assessment • Speech enhancement experience, noise suppression, dereverberation, speaker separation • Experience with audio-visual ML models, multi-modal learning • Familiarity with hearing science metrics, HASQI, HASPI, PESQ, POLQA • Signal processing fundamentals, DSP, beamforming, acoustic measurements • Experience with human participant research and perceptual data collection • Background in computational hearing science or auditory cognitive neuroscience