What you'll do Design and optimize AI image/video generation pipelines for high-concurrency inference Integrate and fine-tune open-source video generation models (e.g. Wan2.2, CogVideoX) Optimize GPU inference performance using TensorRT, xformers, and quantization techniques Implement LoRA-based fine-tuning for character consistency and style customization Collaborate with ML researchers to accelerate the path from research to production Requirements 5+ years of AI/ML engineering experience Hands-on experience with video or image generation systems in production Strong PyTorch skills; experience with Diffusers and DiT architectures Solid understanding of GPU inference optimization (TensorRT, xformers, quantization) Practical LoRA fine-tuning experience, including hyperparameter tuning and overfitting control Ability to independently build and deploy inference API services (Python / Node.js)