Machine Learning Engineer - Audio / Speech

Summary:

• Own and sustain a family of production machine learning models.
• Day to day responsibilities include: maintain ML models’ inference services and evaluation pipelines, integrate models into internal tools, and support the users and tooling owners using models.
• Tech stack: Python, PyTorch, Bento (Jupyter-style notebooks), Client's internal model-serving and always-on inference capacity, REST/GraphQL-style endpoints, and a lightweight web UI.

Responsibilities:

• Own a family of deep-learning models end to end:

architecture, checkpoints, evaluation pipelines, serving infrastructure, and failure modes

• Integrate these models into internal and XFN tools and workflows via API/endpoint integration and web UI onboarding.
• Operate always-on model inference capacity:

monitor traffic, resolve throttling, tune auto-scaling, request additional capacity, redeploy, and escalate to platform owners as needed

• Run analysis and interpret model evaluations on request, apply minor bug fixes and preprocessing changes, and manage version bumps and checkpoint swaps
• Communicate with and support model users and tooling owners across various domains including audio engineers, SDEs, research scientists, TPMs etc.
• Serve as oncall for the covered services.

Minimum Qualifications:

• Bachelor's degree in Computer Science, Electrical Engineering, or a related technical field, or equivalent practical experience.
• Proficiency in Python and a deep-learning framework such as PyTorch.
• Knowledge of Machine Learning concepts and ML engineering practices.
• Basic knowledge of audio and signal processing.
• Ability to work independently

Preferred Qualifications:

• Master's or PhD degree in Electrical Engineering, Audio Engineering, Speech or Signal Processing, Acoustics, Computer Science, or a related technical field.
• 2+ years of hands-on experience deploying and maintaining machine learning models in production. Experience operating production services, including oncall, ticket queues, runbooks, access management, and escalation
• Working familiarity with audio concepts (waveforms, sample rate, spectrograms) sufficient to sanity-check model outputs
• Excellent communication skills with nonML audience, including audio engineers and scientists.
• Experience with Client internal ML platform tooling stack.
• Experience with audio, speech, or perceptual quality models (e.g. MOS prediction)
• Experience developing lightweight web front ends

Top 3 must-have HARD skills:

• Proficiency in Python and a deep-learning framework such as PyTorch.
• Knowledge of Machine Learning concepts and ML engineering practices.
• Basic knowledge of audio and signal processing.

Good to have skills:

• Experience with audio, speech, or perceptual quality models (e.g. MOS prediction)
• Working familiarity with audio concepts (waveforms, sample rate, spectrograms) sufficient to sanity-check model outputs
• Experience with Client’s internal ML platform tooling stack.