Research Scientist

Cantina is a social AI company building advanced real-time generative models that bring characters and personalities to life. Their flagship platform enables new forms of storytelling and creative expression, with plans to expand into broader ecosystems. The company is growing its Singapore team and seeking a Research Scientist to lead foundational work on video generation models.

About Cantina

Cantina develops cutting-edge social AI technology focused on generating realistic, expressive video and character models. The company operates at the intersection of generative AI and creative tools, serving creators and platforms that depend on high-quality, real-time media generation. As the AI job market expands rapidly—with AI/ML architect roles growing 196.5% year over year—Cantina is investing in research talent to advance the core science behind video generation and model efficiency.

What you'll do

• Lead research initiatives on video generation models, managing the complete research pipeline from conception through validation and implementation

• Build and operate large-scale data systems for ingesting, preprocessing, and delivering video training data, including orchestration, scheduling, monitoring, and failure recovery

• Design distributed data pipelines using frameworks like PySpark or Ray, with infrastructure managed through Kubernetes or similar orchestration platforms

• Optimize cloud data storage and transfer across AWS, GCS, or Azure, balancing cost, throughput, and operational needs

• Research and develop distillation approaches for diffusion and flow-based video models, including guidance and adversarial distillation methods that preserve generation quality while reducing inference requirements

• Build reward models and preference-based fine-tuning systems aligned with human judgments on aesthetics, motion quality, and prompt adherence

• Partner with data, infrastructure, and modeling teams to translate research outcomes into durable model improvements

What you'll bring

• Hands-on experience building or scaling data systems and pipelines for machine learning at significant scale

• Proficiency with distributed processing frameworks, container orchestration (Docker, Kubernetes), and cloud platforms (AWS, GCS, Azure)

• Strong background in post-training methods for large-scale diffusion or flow models, with demonstrated expertise in distillation, inference efficiency, and quality preservation

• Experience with reward modeling or preference-based fine-tuning methods such as RLHF or DPO

• Solid understanding of video and media processing tools, multimodal data, and the relationship between pretraining and post-training outcomes

• Fluency in Python with PyTorch or JAX, plus a track record of driving independent research projects from conception through validation

Nice to have

• Publications at top venues such as NeurIPS, ICML, ICLR, CVPR, ICCV, or ECCV

What they offer

• Competitive salary and company equity

• Personal time off, paid holidays, and health insurance

• Global travel insurance for international trips

• Monthly equipment stipend of S$635

• Full home office equipment provided