AI/ML Engineer (Fine-Tuning CogVideoX 5B for Video Inpainting) - Part-time - Worldwide

Job Title: Senior AI/ML Engineer for 3D Movie Conversion (Freelance)
Project: Fine-Tuning AI Model for Image Inpainting in 3D Movie Conversion

Role Overview:

Redmond seeks a skilled AI/ML engineer with expertise in computer vision and video processing to fine-tune an open-source model for image inpainting, specifically to support 2D-to-3D movie conversion. The ideal candidate should possess a strong understanding of video processing, deep learning frameworks, and model optimization techniques. Your role will involve customizing the model to enhance its performance for specific video datasets. If you have a passion for improving AI models and a background in video inpainting, we would love to hear from you!

We have a dataset of 40k video clips. Each clip has a rgb, mask, and gt image sequence. The model should be designed to take the rgb and mask as inputs and trained to reproduce the GT.

Responsibilities:

• Fine-tune the CogVideoX 5B or a comparable open-source model for image inpainting tasks
• Customize the model to handle 2D movie frames, creating 3D visual depth by filling warped areas
• Use Redmond's dataset (40,000 video clips with RGB, mask, and ground truth sequences) to train and validate model performance
• Implement model adjustments for high accuracy in generating inpainted images based on RGB and mask inputs to match the ground truth

Required Skills:

• Proficiency with PyTorch and strong expertise in machine learning and computer vision
• Experience with video processing and model fine-tuning for image and video-based applications
• Proven track record of optimizing non-LLM models, ideally in the image or video domain