5 days ago
Base Salary
$200k - $350k/yr
Responsibilities
- Build and operate performant, scalable infrastructure for acquiring, ingesting, and combining massive text datasets.
- Design and operate reproducible data pipelines for preprocessing, filtering, deduplication, and augmentation.
- Run ablation experiments to evaluate how data sources, processing choices, and mixture weights affect model quality.
- Partner with research and infrastructure teams on data loading, versioning, and experimentation pipelines.
- Establish data-quality standards and connect dataset characteristics with model behavior.
- Source novel datasets and manage relationships and budgets with external data vendors and partners.
Requirements
- Hands-on experience with ML data infrastructure, including training-data pipelines, dataset versioning, large-scale data loading, and the relationship between data systems and model training and inference.
- Strong modern engineering execution, including clean, well-tested code and the ability to select appropriate tools for a problem.
- Familiarity with building and evaluating datasets for generative models and a working knowledge of model training and inference.
- Experience with large-scale parallel data processing using Ray, Spark, or Kubernetes is a preferred qualification.
- Experience with language-model pretraining is a preferred qualification.
Benefits
- In-person work from offices in San Francisco, London, or Bangalore.
- Visa sponsorship support is assessed case by case and depends on the role and location.
- US employees receive fully covered medical, dental, and vision insurance for themselves and their families.
- US employees receive 9 weeks of paternity leave and 12 weeks of maternity leave.
- US employees have access to a 401(k), monthly commuter allowance, flexible PTO, daily meals and snacks, and an equity package.
Tech Stack
Categories
Data Engineering
About Cartesia
Cartesia builds real-time audio AI models and APIs for developers creating interactive voice agents and applications. Its products include Sonic for low-latency speech recognition, text-to-speech, and speech-to-speech, and Ink as companion tooling for interactive agents. Founded in 2023, the company is privately held and offers its technology as an AI/API platform for software teams.