Apple AR/VR Job | Machine Learning Project Lead (Body Technologies) m/f/d
Citys（岗位城市）: Munich, Bavaria-Bayern, Germany
Are you eager to work at the intersection of Computer Vision and Deep Learning, on projects that turn technical innovation into Apple products which touch the lives of millions of people? Then join the VCV Body Technologies team at the Munich Vision Lab as ML tech lead (f/m/d). Our team has brought to live the Persona technology that powers FaceTime on Apple Vision Pro. Help us develop ground breaking technology for more human body understanding use cases.
We are looking for candidates with outstanding technical expertise, with a track record in algorithm development for image-based shape and mesh recovery, or a highly related disciplines such as human motion capture. Experience with neural rendering or AI image generation is a plus. You should be able to quickly prototype different algorithm solutions, but also to elevate them to production standard.
Join us for the rare opportunity to work on computer-vision-driven products that go beyond the state of the art and that delight and inspire millions of Apple’s customers every single day.
Strong theoretical background + practical experience in Deep Learning with proficiency in PyTorch or Tensorflow
Comprehensive knowledge in 3D Computer Vision including image formation and multi-view geometry
Solid foundational and applied math knowledge, particularly around Linear Algebra and Optimization.
Great Python skills for writing efficient and maintainable solutions in larger code bases
Track record in parametric shape estimation, mesh recovery or highly related discipline
Strong communication, technical planning and decision-making skills
Resilience to uncertain and complex environments
Experience with neural rendering or AI image generation is a plus
Proficiently in English
As a machine learning tech lead, your tasks within our team encompass the entire project scope, from the initial cross-functional definition, large-scale ground truth data generation to prototype development, product integration and quality evaluation.
As ML algorithms team, we are responsible to source our training data by specifying requirements and ensuring their implementation in both real and synthetic data campaigns. Proficiency in data handling, including capturing, processing and managing large datasets, data cleaning, transformation, and augmentation, is essential to tailor the data to specific model and use case needs.
Identifying the ideal algorithm and model architecture for a particular use case and hardware configuration demands a comprehensive knowledge of available options. Therefore, it is a valuable asset to have experience with a range of deep learning techniques, such as traditional CNNs, transformers and, for instance, neural rendering or diffusion methods. Numerous data-driven decisions must be made, necessitating expertise in model evaluation using traditional or custom KPI metrics as well as computational efficiency on large GPU clusters (training) and on-device SOCs (deployment).
Building image-based 3D reconstructions requires a strong understanding of 3D computer vision and image formation principles, as well as mathematics, including linear algebra and optimization. Experience with 3D data representations such as e.g. meshes, point clouds, depth textures, or voxels is advantageous for handling complex 3D data.
Excellent Python programming skills and a deep understanding of best software engineering practices will allow you to make a meaningful impact on a large-scale project. GPU programming could further accelerate training.
In this role effective communication is essential for conveying requirements, challenges, and solutions to a diverse audience, both technical and non-technical, within cross-functional teams. Strong communication skills and a collaborative mindset are a must.