We are seeking a highly skilled Generative AI Engineer to join our team and drive innovation in image and video restoration. You will work on state-of-the-art algorithms for deblurring, denoising, super-resolution, and related tasks, leveraging generative models such as diffusion models, GANs, and transformers. This role requires both strong research ability and practical engineering skills to develop scalable solutions for real-world applications.
Description
- Research, design, and implement generative AI models for image and video restoration (e.g., deblurring, denoising, super-resolution, inpainting, frame interpolation).
- Build and optimize training pipelines for large-scale datasets, including preprocessing, augmentation, and distributed training.
- Evaluate restoration performance using both objective metrics (PSNR, SSIM, LPIPS) and subjective/perceptual quality measures.
- Develop scalable and efficient inference pipelines, optimizing for latency, throughput, and memory.
- Stay current with the latest research in computer vision and generative AI, and translate novel ideas into practical solutions.
- Collaborate with cross-functional teams to integrate restoration models into production systems.
Preferred Qualifications
Track record of publications or open-source contributions in generative AI, computer vision, or image/video restoration.
Experience with real-world video data processing (e.g., raw domain, HDR pipelines, ISP sharpening).
Familiarity with cloud-based large-scale dataset management (e.g., S3, distributed file systems).
Experience with real-time or near-real-time video restoration and performance optimization.
Knowledge of advanced motion analysis (scene change detection, temporal consistency checks, optical flow with RAFT/PWC-Net).
Familiarity with deployment on diverse hardware (edge devices, mobile, GPU acceleration).
Practical experience with efficient model deployment, including model compression, quantization, distillation, and hardware optimization (e.g. TensorRT, mixed precision).
Minimum Qualifications
Master’s or Ph.D. in Computer Science, Electrical Engineering, Applied Mathematics, or a related field.
Strong background in computer vision and deep learning, with proven experience in generative models (diffusion, GANs, transformers).
Proficiency in Python and deep learning frameworks (PyTorch preferred).
Experience with large-scale image/video datasets and distributed training (multi-GPU or multi-node).
Solid understanding of image and video restoration metrics (PSNR, SSIM, LPIPS) and perceptual evaluation.
Hands-on experience with video preprocessing, such as motion estimation and optical flow, frame alignment and stabilization, temporal consistency techniques, and video encoding/decoding.
Strong software engineering skills: clean code, Git, debugging, optimization.
