Research Engineer - Vison Foundation Models

About Netropy AI

Netropy AI is a startup pioneering Physical AI and high-fidelity simulation solutions for industries such as autonomous driving and robotics. We leverage video-based 3D reconstruction, advanced computer vision, and generative technologies to bridge the gap between the virtual and physical worlds.

About the Role

We are looking for a Research Engineer to join our team and push the boundaries of Physical AI. In this role, you will combine cutting-edge research in generative AI and foundation models with image and 3D reconstruction. You will play a pivotal role in developing the next generation of simulation technology.

• Generate, propose, and evaluate novel research ideas to drive the continuous evolution of our AI technology stack.
• Develop and implement state-of-the-art 3D/4D reconstruction technologies, including Gaussian Splatting, Vision Generative Models, and Geometric Foundation Models.
• Manage the lifecycle of large-scale datasets, including preprocessing and curation, to support robust model training and evaluation.
• Optimize research prototypes and integrate developed AI models into our core products, ensuring seamless performance and scalability.

To be considered for this role, applicants must meet the following criteria

• 2+ years of research experience in Computer Vision, Computer Graphics, or related fields (Academic or Industrial).
• A deep understanding of Vision Generative and Foundation Models (e.g., Diffusion models, Transformers).
• Strong grasp of traditional computer vision and fundamental mathematics (Linear Algebra, Calculus, Statistics, and Probability).
• Extensive knowledge of Deep Learning literature with the ability to rapidly comprehend and implement new technologies.
• Strong critical thinking skills, a hands-on approach to problem-solving, and the ability to learn quickly in a fast-paced environment.
• Professional proficiency in English for technical collaboration.
• Experience researching Vision Diffusion or Flow Matching models.
• Large-Scale Training: Hands-on experience training or fine-tuning Large Vision Models (LVMs) on multi-GPU setups.
• Publication Record: A track record of publishing research at top-tier conferences such as CVPR, SIGGRAPH, ICCV, NeurIPS, or ECCV.
• Great Accessibility: Located a 5-minute walk from Nambu Bus Terminal Station (Line 3).
• Flexible Work System: Flexible arrival time (up to 11:00 AM).
• Meal Allowance: Lunch provided to support team well-being.
• Latest Equipment: High-end equipment and tools provided to maximize productivity.