🛠 Tech Stack

💼 About This Role

You'll implement and train state-of-the-art image and video generation models at Cantina, a social AI company. You'll own the engineering execution that turns research ideas into working systems, building data pipelines and scaling training. This is a high-impact role focusing on large-scale diffusion models and distributed training.

🎯 What You'll Do

• Build and maintain end-to-end data pipelines for large-scale image and video datasets.
• Implement model architectures and maintain high-throughput PyTorch training loops.
• Run large-scale training experiments and debug training instabilities.
• Apply quantization and distillation to compress models.
• Collaborate with researchers to implement state-of-the-art papers.

📋 Requirements

• 2–5 years of hands-on experience building and training ML systems
• Fluency in PyTorch for both training and inference
• Experience training or fine-tuning diffusion models
• Ability to read and implement AI research papers in computer vision

✨ Nice to Have

• Experience with distributed training (DDP, FSDP, DeepSpeed)
• Familiarity with inference optimization (torch.compile, TensorRT)

🎁 Benefits & Perks

• 💰 Competitive salary and equity
• 🏥 99.99% covered medical, dental, vision
• 🌴 42 days PTO (15 PTO, 10 sick, 15 holidays, 2 floating)
• 👶 Generous parental leave & fertility support
• 🧘 $500/month lifestyle spending account