Machine Learning Researcher - Audio

2d2 days ago

Protege

Remote · Full-time · $160,000 – $220,000

About this role

We are building Protege to solve the biggest unmet need in AI — getting access to the right training data. The platform facilitates the secure, efficient, and privacy-centric exchange of AI training data for speech and audio models.

We’re seeking a Machine Learning Researcher focused on audio data quality, ML data evaluation, and quality control. You will lead the evaluation and optimization of large-scale speech datasets used to train audio, speech, and multimodal models.

This role involves original research and method development: designing new approaches for measuring audio data quality and validating them against downstream model outcomes. You will translate research insights into practical evaluation tools, filtering rules, and quality standards across the data platform.

We’re a lean, fast-moving, high-trust team of builders obsessed with velocity and impact. Our culture is built for people who thrive on ambiguity, own outcomes, and want to shape the future of data and AI.

Requirements

Deep interest in audio data quality and signal understanding, with a focus on how acoustic degradation affects ML training.
Experience applying existing audio quality metrics and developing new methods for evaluating speech datasets.
Comfortable operating in both research and hands-on implementation modes.
Strong understanding of acoustic properties such as effective bandwidth, spectral energy distribution, and distortion.
Ability to design and validate evaluation frameworks that predict downstream model performance.
Experience working with large-scale speech datasets and familiarity with ASR, TTS, or speaker modeling systems.
Proficiency in Python and machine learning frameworks for audio processing and analysis.

Responsibilities

Research how audio quality, signal properties, and dataset composition affect downstream model training and deployment.
Develop new metrics, benchmarks, diagnostics, and evaluation frameworks for measuring audio data quality that predict ML model performance.
Analyze Protege’s audio catalog and maintain clear quality scorecards and metrics for key speech datasets.
Develop methods to measure true acoustic properties directly from the waveform, including bandwidth, noise, clipping, reverberation, and codec artifacts.
Build workflows for segment-level quality evaluation that surface localized degradation in diarized or segmented speech regions.
Apply multiple complementary quality metrics to detect bandwidth mismatches, resampling artifacts, and other forms of degradation.
Design and run evaluations linking data quality to downstream model performance for ASR, TTS, speaker modeling, and multimodal systems.