Cantina · Singapore
Data & AIResearchFullTimeposted 3mo ago
About Cantina:
Cantina Labs is a social AI company, developing a suite of advanced real-time models that push the boundaries of expression, personality, and realism. We bring characters to life, transforming how people tell stories, connect, and create. We build and power ecosystems. Cantina, our flagship social AI platform, is just the beginning.
About the Role:
Cantina is expanding, and we're looking for an ML Engineer to join our growing Singapore team! In this role, you will build and scale systems for ingesting, processing, and delivering large-scale video and multimodal data for model training. You'll own the full pipeline — from raw content to curated, filtered, and training-ready datasets — with a focus on speed, reliability, reproducibility, and cost-efficiency. You'll partner closely with curation and modeling teams to operationalize evolving dataset recipes and iterate on approaches that improve model outcomes.
What You’ll Do:
Design and scale distributed data pipelines for preprocessing, dataset generation, and repeated dataset refreshes
Own workflow orchestration, job scheduling, monitoring, and failure recovery for large-scale data processing jobs
Implement and maintain containerized pipeline infrastructure using Kubernetes or equivalent orchestration systems
Optimize cloud-based data storage and movement across providers (AWS, GCS, or Azure) for cost, throughput, and operational efficiency
Define and implement best practices for dataset storage layout, versioning, caching, retention, and access patterns
Design and implement curation pipelines that determine which video and image content is selected, filtered, and retained for model training, including image-text pair datasets used in joint training regimes
Build and improve VLM-based captioning and metadata generation workflows at scale across both video and image data
Develop and apply quality and aesthetic scoring models, CLIP-based semantic filtering, and other signal-extraction approaches for data selection
Build tooling to support deduplication workflows at scale, including near-dedup and exact deduplication pipelines over large video corpora
Analyze dataset composition, identify quality issues, and iterate on curation logic to improve training outcomes
Define and evolve standards for what constitutes high-quality, training-ready video data across different training regimes
What You’ll Bring:
Strong hands-on experience building or scaling large-scale data systems and pipelines for machine learning, including dataset curation, filtering, and quality improvement
Experience with distributed data processing frameworks such as PySpark or Ray, and orchestration tools such as Airflow or equivalent
Familiarity with containerization and container orchestration, including Docker and Kubernetes
Experience working with cloud-based data storage and compute (AWS, GCS, and/or Azure), including tradeoffs around cost, throughput, storage layout, and access patterns
Experience with VLM-based captioning pipelines or quality/aesthetic scoring models for video or image data, including curation of image-text pair datasets for joint image-video training
Familiarity with CLIP-based or embedding-based filtering and semantic data selection techniques
Familiarity with video and media processing tools such as FFmpeg, PyAV, DALI, or OpenCV, and relevant libraries such as Decord, torchvision, PyTorchVideo, or torchaudio
Proficiency in Python
Strong problem-solving, communication, and documentation skills
Benefits We Offer:
Competitive salary and generous company equity
Personal time off and paid holidays
Health insurance
Global travel insurance: Covers you when traveling internationally
Monthly spending stipend: $500 (~S$635)
Equipment: All equipment needed for your home office