About the Role :

You will design and own scalable speech and audio processing systems including :

- Audio ingestion - preprocessing - tagging - QA - dataset delivery pipelines

- Automated audio quality checks (SNR, clipping, silence detection, noise profiling)

- ASR integration & evaluation (Whisper, Kaldi, SpeechBrain, etc.)

- WER / CER computation and optimization

- Transcription validation workflows

- Audio normalization, resampling, segmentation, chunking

- Structuring enterprise-grade speech datasets

- Speaker verification, dialect tagging, emotion tagging

Candidates must demonstrate hands-on ownership of speech pipeline components.

Required Experience :

- 4 to 7 years of hands-on experience in speech/audio processing

- Strong Python

- Experience with Librosa, torchaudio, PyDub, SpeechBrain, Kaldi or similar

- Experience working with ASR systems in production

- Strong understanding of sampling rates, spectrograms, MFCCs, preprocessing

- Experience computing and improving WER/CER

- Experience building pipelines on AWS / GCP / Azure

Strongly Preferred :

- Strong academic foundation in signal processing, speech technology, or related field preferred

- Experience fine-tuning ASR models

- Speaker diarization

- Multilingual speech systems

- Large-scale dataset QA automation

Not a Fit If :

- Your experience is primarily prompt engineering

- You have only wrapped Whisper APIs without audio-level ownership

- You are a generic full-stack or RAG engineer