Senior Machine Learning Engineer- Speak
|Published||December 16, 2022|
|Location||Finland (Remote), Finland|
The Personalization team makes deciding what to play next easier and more enjoyable for every listener. From Blend to Discover Weekly, we’re behind some of Spotify’s most-loved features. We built them by understanding the world of music and podcasts better than anyone else. Join us and you’ll keep millions of users listening by making great recommendations to each and every one of them. We ask that our team members be physically located in Central European time or Eastern Standard/Daylight time zones for the purposes of our collaboration hours.
Within Personalization, the Speak product area crafts voice models that match human-level emotional ability, so we can deeply engage our listeners and support creators, at scale. Our groundbreaking work on speech synthesis relies on state-of-the-art deep learning methods and evaluation techniques, highly efficient data processing and model serving, and capturing audio of outstanding quality from our voice actors.
We’re looking for senior machine learning (ML) engineers with experience across the whole model development lifecycle. You’ll be deeply involved in creating our next generation of larger speech synthesis models. Central to this effort will be scaling up our dataset creation and model training systems to improve our experimental agility. You'll also play a key role in improving our evaluation pipelines to ensure that the quality of our voices delight Spotify users.
What You'll Do
- Develop scalable pipelines for data preparation, model evaluation, and training of large models for speech synthesis.
- Collaborate with our research team to raise engineering standards and to design, build, evaluate, and refine our models.
- Contribute to building and improving our infrastructure for research, development, and model serving.
- Share your knowledge and experience of how to operate and scale ML systems.
- Be part of an engineering team at Speak dedicated to building and serving our models at scale, as well as an active group of machine learning practitioners in the Personalization mission and across Spotify.
Who You Are
- You have a strong background in ML with experience in a generative Modeling domain, ideally in speech synthesis or natural language processing. Knowledge of audio processing is a bonus.
- You have worked with the latest model architectures in generative modeling such as transformers, generative adversarial networks (GANs), diffusion models, normalizing flows, or variational autoencoders (VAEs).
- You have trained and built large ML models at scale using PyTorch, as well as experiment tracking systems such as MLFlow, and distributed training frameworks such as Horovod, DeepSpeed, Ray, FairScale, or torch distributed.
- You have used techniques to search hyperparameter spaces, to find better neural architectures, and to reduce model size, cost, and latency for production usage.
- You have hands-on experience of developing production machine learning systems at scale on the cloud in Python, Java, or similar languages.
- You appreciate the value of high quality data collection and evaluation processes, always looking for ways to scale up and automate.
- You have a strong experimental mindset with a track record of interpreting and implementing methods described in research papers.
- You advocate for agile software processes, data-driven development, the production of secure and reliable systems, model provenance, and principled experimentation.
Where You'll Be
- We are a distributed workforce enabling our band members to find a work mode that is best for them!
- Where in the world? For this role, it can be within the EMEA region in which we have a work location and is within working hours.
- Working hours? We operate within the Central European and GMT time zones for collaboration and ask that all be located in that time zone.
- Prefer an office to work from home instead? Not a problem! We have plenty of options for your working preferences. Find more information about our Work From Anywhere options here.
Spotify is an equal opportunity employer. You are welcome at Spotify for who you are, no matter where you come from, what you look like, or what’s playing in your headphones. Our platform is for everyone, and so is our workplace. The more voices we have represented and amplified in our business, the more we will all thrive, contribute, and be forward-thinking! So bring us your personal experience, your perspectives, and your background. It’s in our differences that we will find the power to keep revolutionizing the way the world listens.
Spotify transformed music listening forever when we launched in 2008. Our mission is to unlock the potential of human creativity by giving a million creative artists the opportunity to live off their art and billions of fans the chance to enjoy and be passionate about these creators. Everything we do is driven by our love for music and podcasting. Today, we are the world’s most popular audio streaming subscription service.