Responsibilities
- Work with founders at Pebblely to build and deploy cutting-edge AI research with generative image models
- Candidates will work directly with the technical co-founder, who has published with researchers from Stanford, PyTorch, and OpenAI
- Specifically, you will explore the following research directions relating to image generation:
- Algorithmically evaluating realism and aesthetic qualities of generated images
- “Over”-painting (i.e. adding appropriate shadows, highlights, or other effects on existing product images, to better integrate it into the inpainted background)
- Object replacement (replacing an object in an existing image with another object)
- Reduction of artifacts from image inpainting (i.e. reduce unwanted extensions to a product, such as wheels or handles)
Requirements
- Proficiency in Python and PyTorch
- Ability to understand, write and modify detailed PyTorch code, beyond running high-level scripts and models
- Prior projects illustrating the skills above, which are available for review
Good to have
- Understanding of diffusion and attention models
- Experience with using and modifying the diffusers library
- Experience with writing and deploying Dockerfiles
