16-20 Weeks

Role Overview:

The MLOps Audio Engineer is a hands-on practitioner embedded within audio delivery function. You will serve as the technical audio expert on the team, translating customer requirements into recording configurations, processing pipelines, QA criteria, and evaluation frameworks.

This is not a research role. You will operate across live production programs, collaborate with engineering and operations teams to build and maintain robust pipelines, and diagnose audio-quality issues at scale.

Key Responsibilities

Audio Data Pipeline Design

- Design end-to-end workflows for audio and speech data collection, processing, annotation, QA, and delivery, from recording configuration through final dataset output and SPL.

- Translate customer audio requirements into technical specifications, including formats, codecs, sample rates, bit depth, channel configurations, and processing requirements.

- Identify technical risks and feasibility gaps before projects move into scaled production.

MLOps, Automation, and Production Support

- Build and maintain pipelines for automated audio ingestion, processing, transcription, diarization, quality checks, and output generation.

- Integrate third-party speech APIs and open-source models into production workflows.

- Collaborate with engineering teams to automate repetitive technical checks and reduce manual QA requirements.

- Monitor pipeline performance, data quality, processing efficiency, and failure rates at scale.

- Drive continuous improvement across audio production workflows.

- Partner with Delivery, Product, Engineering, and Operations teams during project design and launch.

- Convert learnings from individual programs into reusable audio standards, tools, and playbooks.

ASR, Diarization, and Transcription Systems





- Evaluate and benchmark ASR models and APIs, including Whisper, NVIDIA NeMo, Deepgram, AssemblyAI, Speechmatics, Google Speech, Azure Speech, AWS Speech, and similar systems.

- Assess performance across languages, accents, recording environments, and speaker conditions.

- Define evaluation methodologies using WER, CER, confidence scores, and human-validation protocols.

- Recommend ASR solutions based on accuracy, language coverage, latency, cost, and deployment constraints.

- Design and evaluate speaker-diarization pipelines covering speaker segmentation, turn detection, overlapping speech, and speaker attribution.

- Apply DER and related diarization metrics.

- Design human-in-the-loop correction and validation workflows for machine-generated transcripts at scale.

Audio Quality and Signal Analysis

- Define and evaluate audio-quality parameters, including SNR, clipping, distortion, loudness, background noise, echo, reverberation, packet loss, cross-talk, overlapping speech, sampling rate, and codec artifacts.

- Develop automated and human QA criteria for determining whether recordings meet customer acceptance thresholds.

- Build tooling for batch audio analysis, dataset validation, and signal-quality reporting across large audio corpora.

- Lead root-cause analysis when audio-quality issues arise during production.

- Diagnose whether issues originate from the environment, device, pipeline, or configuration.





Multilingual and Multi-Speaker Programs

- Evaluate technical challenges across languages, accents, dialects, and code-switching scenarios.

- Determine language-specific ASR and diarization requirements for single-speaker, multi-speaker, conversational, and multilingual programs.

- Account for environmental, device, and speaker variability.

- Recommend approaches for single-track versus multi-track recordings and speaker-level audio capture in multi-speaker environments.

- Work with language experts and operations teams to establish appropriate quality standards for linguistically diverse datasets.

- Identify languages or recording conditions that require specialized models or elevated human review.

Requirements

- 3+ years of hands-on experience in speech/audio engineering, speech AI, ML engineering, MLOps, or a directly related field.

- Robust fundamentals in digital audio, including SNR, clipping, gain, sampling rate, bit depth, channel configurations, codecs, noise, echo, and reverberation.

- Practical experience evaluating, benchmarking, and integrating ASR systems into real-world production workflows.

- Working knowledge of speaker diarization and multi-speaker conversational audio.

- Experience working with real-world noisy, uncontrolled, or variable-quality audio data rather than only clean studio recordings.

- Strong Python skills with hands-on experience using audio-processing libraries and pipeline tooling.

- Ability to analyze large audio datasets programmatically and identify quality issues at scale.

- Clear communication skills, with the ability to translate technical findings into actionable recommendations for non-technical operations and delivery teams.

📌 MLOps Audio Engineer (India)
🏢 Turing
📍 India