AI Research Engineer (Post-Training)

About the Role

We are advancing multimodal AI technologies aimed at creative content creation involving images and 2D animations. Your role involves adapting, evaluating, and deploying foundational visual, multimodal, and generative AI models to enhance image creation and manipulation, visual comprehension, consistent styling and character representation, animation support, and quality control of creative output.

Responsibilities

• Assess and choose visual, multimodal, and image-generation foundational models; set quantifiable benchmarks and development plans.
• Execute supervised fine-tuning, LoRA/PEFT, preference optimization, continual adaptation, distillation, and other post-training techniques with proprietary datasets.
• Create solutions for image creation and editing, visual data interpretation, automatic tagging, style identification, character consistency checks, asset quality evaluation, and detection of near-duplicate items.
• Develop 2D animation-related features including understanding parts, aiding layer extraction, estimating poses and skeletons, assisting rigging, transferring motion, generating animations, and assessing quality.
• Integrate model results with structured constraints, validation methods, safety protocols, and human review processes to increase reliability and expand editability and publishability.
• Construct systems for offline and human-involved evaluation to gauge generation quality, editing success, style alignment, initial acceptance, user satisfaction, and measure latency and inference costs.
• Collaborate with AI data engineering to convert model errors, artist corrections, user feedback, and moderation data into effective training and evaluation resources.
• Partner with product, design, and engineering teams to provide production-level services, batch processing, and creator tools.
• Ensure all training, evaluation, inference, monitoring, and model versioning workflows are replicable and maintainable.

Qualifications

• At least one year of experience in machine learning, computer vision, multimodal AI, generative AI, or applied AI research environments.
• Proficient in PyTorch with skills in model training, fine-tuning, evaluation, deployment, or optimization of inference.
• Practical experience with methods such as supervised fine-tuning (SFT), LoRA/PEFT, DPO/RLHF/RLAIF, reward modeling, model evaluation, or continuous learning.
• Expertise in at least one of image understanding, image generation/editing, segmentation, pose estimation, diffusion modeling, vision-language integration, or video/animation generation.
• Capability to convert research prototypes into stable, observable, and maintainable production implementations.
• Excellent skills in experimental design, data analysis, and cross-team collaboration.

Preferred Skills

• Experience developing AI products for image assets, design software, advertising content, game art, or creator platforms.
• Knowledge of 2D character animation workflows such as Spine or Live2D, game asset management, computer graphics, rigging, or kinematics.
• Understanding of structured generation techniques, JSON/schema constraints, geometric constraints, content safety measures, or graphic asset validation.
• Expertise in distributed training, GPU performance optimization, quantization, accelerating inference, or reducing computational costs.
• History of shipping generative AI solutions, publishing research papers, or contributing to open-source codebases.