Senior Software Engineer — Streaming Speech Translation Data

AppleCupertino, CaliforniaOn-siteFull-timeSenior, 5–8 yearsListed 17 minutes ago

Apply now

About this role

We are looking for a Senior Software Engineer to architect, build, and scale our end-to-end streaming speech translation data pipelines. In this role, you will own the infrastructure and data workflows that transform raw multilingual data into high-quality training and evaluation datasets for real-time speech-to-speech and speech-to-text foundation models.

You will work at the intersection of distributed systems, large-scale data engineering, and generative AI—integrating LLM APIs, forced alignment, text normalization, and GPU-accelerated Text-to-Speech (TTS) synthesis into resilient, high-throughput pipelines.

You will be part of a team that's responsible for a wide variety of language technologies related development activities. Your focus will be on developing the model automation pipelines which are highly scalable, robust and efficient. The role will be part of the model automation team to deal with large quantities of data, apply the state-of-the-art methods in deep learning to tackle real world problems, create the production quality models at scale and set up ML CI/CD pipelines.

Minimum Qualifications

Experience:5+ years of industry experience building large-scale distributed data pipelines, backend systems, or ML infrastructure.

Distributed Computing:Strong proficiency in Python and hands-on experience with distributed processing frameworks and modern data lakehouse formats.
GPU & Model Serving Integration: Experience orchestrating GPU-accelerated workloads and managing high-concurrency API calls with robust rate-limiting, retry, and batching strategies.
Speech/NLP Fundamentals:Practical understanding of audio data processing, forced alignment, and multilingual text processing pipelines.
Engineering Rigor:Proven track record of owning production systems end-to-end, including monitoring, automated testing, CI/CD, and debugging complex distributed bottlenecks.

Preferred Qualifications

Direct experience with speech-to-speech translation, streaming ASR/TTS, or forced alignment.

Familiarity with text normalization and inverse text normalization across diverse languages and scripts.

Experience optimizing GPU utilization for batch audio synthesis and inference.

Background in building automated data quality scoring and filtering pipelines for speech or text corpora.