AMI Scientist/Interns - Geometry and 3D Vision

Job description

# About AMI We are building a new breed of AI systems that (1) understand the real world, (2) have persistent memory, (3) can reason and plan, and (4) are controllable and safe. We are a team of scientists and engineers building frontier world model-based AI. We combine the scientific rigor of a top-tier research institute with focus on engineering excellence and execution. We are a global company, with offices in Paris, Montreal, New York, and Singapore. Come build the future of AI with us! # About this Role AMI believes AI agents should predict and plan using an internal model of the world — their world model. We’re looking for new team members to advance the state-of-the-art in world modeling. We believe that video is a rich and abundant source of data reflecting how the world works, and that in general models need to be able to process continuous, high-dimensional data from a variety of sensors to: (a) understand context about the current state of the physical world, (b) make predictions about how the world will evolve, possibly as a result of actions taken, and (c) plan and adapt sequences of actions to complete complex tasks, possibly in dynamic, complex environments. You will work with a team on world model research efforts, with a specific focus on 3D geometry of the real world, including:: - Self-supervised learning methods to efficiently learn from video and other continuous, high-dimensional signals - New architectures that efficiently learn to predict world dynamics from video and other high-dimensional signals - Scalable algorithms for pre-processing and curating video data - Evaluations for benchmarking world model understanding, prediction, and planning - Efficient algorithms for model-based planning and reasoning **Minimum Qualifications:** - Bachelor’s degree or equivalent experience in Computer Science or a related field - Proficiency in Python - Ability to design, run, and analyze experiments independently - Understanding of machine learning fundamentals, large-scale training, and accelerator-based (GPU or TPU) compute environments **Preferred Qualifications:** - Deep expertise in at least one of the following: 3D/4D reconstruction, SLAM or SfM, depth and pose estimation, point tracking. - Hands-on proficiency with a rendering or simulation engine (Blender, Isaac Sim, etc.), including scripted scene generation and batch rendering pipelines. - Demonstrated record of contributing to advanced research projects via publications and/or major model releases - Experience developing evaluation frameworks for world models - Experience releasing and maintaining open-source projects - Proficiency in a deep learning framework (PyTorch or JAX)