About the Role
Develop high-quality character-focused LoRA models for AI applications by leveraging deep learning frameworks and creative design principles.
Responsibilities
- Create detailed character concepts optimized for LoRA training
- Prepare and curate image datasets for model fine-tuning
- Train and refine LoRA models using stable diffusion architectures
- Evaluate output quality across diverse character attributes
- Optimize training pipelines for efficiency and scalability
- Collaborate with researchers to improve model performance
- Document training parameters and results systematically
- Troubleshoot issues in image generation and alignment
- Ensure outputs meet ethical and safety guidelines
- Iterate on models based on feedback and testing
- Maintain version control for trained models
- Support integration of LoRAs into downstream applications
- Stay current with advancements in generative AI
- Participate in peer reviews of model outputs
- Contribute to internal knowledge sharing sessions
- Align character designs with intended use cases
- Apply best practices in data preprocessing
- Monitor for bias or unintended patterns in outputs
- Use metadata tagging to organize model variants
- Assist in developing evaluation benchmarks
- Work with cross-functional teams to define character goals
- Implement feedback loops for continuous improvement
- Test models under varied prompt conditions
- Balance creativity with technical constraints
- Maintain clean and reproducible code environments
Nice to Have
- Experience with 3D pose estimation for character alignment
- Background in animation or game character design
- Published work in generative AI communities
- Contributions to open-source AI projects
- Experience with multi-concept LoRA training
- Knowledge of facial landmark detection tools
- Familiarity with style transfer evaluation metrics
- Prior work with identity-preserving models
- Understanding of copyright implications in training data
- Experience mentoring others in LoRA development
Compensation
Competitive salary with performance-based incentives
Work Arrangement
Hybrid work model with flexible hours
Team
Collaborative AI development team focused on generative models
Technology Stack
- Utilize PyTorch and Hugging Face libraries for model development
- Work with stable diffusion versions 1.5 and XL
- Train models using NVIDIA A100 and H100 GPUs
- Use Weights & Biases for experiment tracking
- Deploy models in containerized environments
Creative Direction
- Focus on expressive, diverse character designs
- Emphasize cultural and stylistic variety
- Prioritize accessibility and representation
- Balance realism with artistic interpretation
- Support both humanoid and stylized forms
Available for qualified international candidates
