Salary: strong>45-50K
Experience requirements: strong>Unlimited experience
Educational requirements: PhD or above
Job Description: strong>
Base: Beijing/Shanghai/Hangzhou/Shenzhen/Dongguan/Xi'an
Job requirements:
1. Doctoral degree in computer, software or related majors
2. Solid foundation, familiar with but not limited to one or more programming languages such as JAVA/C++/Python/JS/HTML/GO, with good programming habits
3. Ability to work independently and solve problems, good at communication, willing to cooperate, enthusiastic about new technologies, good at summarizing and sharing, and enjoy hands-on practice
Note: Based on the doctoral research direction and interests, specific expert exchanges will be matched, including but not limited to neural rendering, 3D vision, depth estimation, image synthesis, computational photography, AI materials, and computer vision
The reference direction is as follows:
[AI Algorithm Application]
1. Master the basic knowledge of artificial intelligence, computer vision, signal processing, image processing, computer software, electronic information, applied mathematics, communication and other related majors
2. Priority will be given to those who possess the following skills and experience:
(1) Familiar with deep learning algorithm models and platform usage, with experience in visual research or publishing papers in international conference journals is preferred
(2) Proficient in Python, C, C++programming experience, solid mathematical foundation, and strong English communication skills
(3) Having experience in prototype development
Intelligent cockpit
1. PhD in computer science, image processing, rendering engines, and related fields
2. Experience in developing map navigation software, AR/3D graphics rendering, virtual robot graphics rendering, etc. is preferred
3. Experience in robotics, autonomous driving, and visual active safety is preferred
4. Proficient in one or more programming languages in Java/C/C++, with cross chip platform and Android/Windows/Linux system development experience preferred
5. Having good teamwork spirit and communication skills
[Autonomous Driving]
1. Responsible for the research and development of cutting-edge technologies such as front fusion perception, geometric measurement, and 3D reconstruction of multimodal sensors such as Camera, LiDAR, and Radar in autonomous driving, ensuring the competitiveness of the company's related products
2. Responsible for exploring breakthroughs and continuous innovation in next-generation intelligent driving technologies such as end-to-end intelligent driving, multimodal large models, and pure visual perception
3. Participate in product requirement specification analysis, collaborate to complete algorithm scheme design, and track the effectiveness improvement of algorithm implementation and application scenarios
Computer Graphics Researcher
1. Research cutting-edge technologies in the field of graphics and images, and conduct pre research and innovation work on technical solutions based on actual business needs; Addressing business pain points and enhancing product user experience through intelligent algorithms; Including but not limited to: exploration and implementation of technologies in neural rendering, 3D vision, depth estimation, image synthesis, computational photography, AI materials, computer vision, and other fields
2. Responsible for insight, innovation, optimization, and implementation of cutting-edge AI algorithms in graphic rendering, to achieve the company's technological competitiveness in the field of neural rendering
3. Insight into the cutting-edge trends of the industry and academia in this field, leading technology planning and competitive forward-looking layout, connecting industry experts, responsible for external influence promotion and internal introduction of new technologies
Media Algorithm
1. Experience in research or development of algorithms such as CV/NLP/recommendation
2. Having a solid foundation in probability theory, information theory, and linear algebra, proficient in various classic models of machine learning and deep learning in fields such as CV/NLP/recommendation, as well as SOTA models and algorithms in corresponding subfields
3. Familiar with deep learning frameworks such as TensorFlow and Pytorch, with the ability to quickly build and optimize models and coding skills
4. Based on technologies such as CV/NLP/recommendation, we have a deep understanding and successful commercial experience in segmented fields such as image classification, image segmentation, object detection/tracking, pose estimation, image generation, image content understanding, visual Q&A, and behavior analysis; Having certain industry influence or serving as a technology leader in related fields is preferred
5. Priority will be given to those who have published papers in top conferences or journals related to CV/NLP/recommendation (CCF A Journal A Conference)
Research on Graphic and Dynamic Algorithm Technology
1. Responsible for researching cutting-edge technologies in the field of graphics and images, and conducting pre research and innovation work on technical solutions based on actual business needs; Addressing business pain points and enhancing product user experience through intelligent algorithms; Including but not limited to: neural rendering, 2D/3D vision, depth estimation, image synthesis, computational photography, etc., technological exploration and implementation in the field of AI materials, and building technological competitiveness in the field of image technology
2. Responsible for insight, innovation, optimization, and implementation of cutting-edge AI algorithms in graphic rendering, focusing on the research and development needs of terminal innovative graphic animation algorithms and scenes. Responsible for patent layout and related technology incubation in business directions including but not limited to automatic generation of bone animations, image data migration animations, and the development and implementation of low-power and high-performance anti saw tooth and post-processing special effects algorithms
3. Responsible for the planning and promotion of related technologies to enhance graphics rendering performance and energy efficiency through the collaboration of software and hardware cores
4. Insight into the cutting-edge trends of the industry and academia in this field, leading technology planning and competitive forward-looking layout, connecting industry experts, responsible for promoting external influence and introducing new technologies internally
Research on Visual/Multimodal Basic Models
Engaged in technical research in the field of computational vision, including but not limited to the following directions:
1. The direction of understanding and generating multimodal large models includes but is not limited to: image/document understanding, image/video generation and editing, object detection, localization, and segmentation. Model architecture, training algorithms, data processing, etc
2. Bottom level vision (image/video super-resolution, denoising, HDR, deblurring, frame interpolation, inpainting, etc.), generative model-based image/video enhancement/restoration/editing, computational photography algorithms, multimodal understanding, etc 3D/4D reconstruction understanding and controllable generation algorithms, domain transfer and small sample learning in visual tasks, neural network/diffsuion model compression&acceleration&quantization, design and development of efficient end-to-end AI models, etc
【 3D Map 】
Targeting autonomous driving, embodied intelligence XR、 In urban design and other scenarios, research 3D object model generation, indoor and outdoor scene layout, realistic rendering and other technologies, build a generative map platform, provide simulation environment data, job responsibilities include but are not limited to:
1. Responsible for the generation of indoor and outdoor 3D scene models, including but not limited to the research and application of related technologies such as 3D geometric model generation, texture&material generation, scene understanding, scene layout, and scene realistic rendering
2. Research on scene knowledge reasoning, 3D model retrieval, 2D/3D layout generation, 3D representation, event representation, text generated 3D, image generated 3D, action generation and other technologies
3. Master the latest technology and trends in the industry, as well as an international technical perspective in the field of vision, lead technological innovation, patent layout, and support the continuous development of various business areas such as autonomous driving and terminals
Research on Visual Algorithms
1. Responsible for the development of key algorithms in industrial vision, including object detection, classification, segmentation, fine object classification, self-learning, unsupervised and other core algorithm development
2. 3D sorting, measurement, and inspection for industry, familiar with commonly used 3D sensor principles, responsible for key algorithms and system development such as 3D point cloud reconstruction, calibration, processing, registration, segmentation, etc
3. Responsible for the development of AI visual algorithm toolchain, including model compression, knowledge distillation, model quantification AutoML、 Image processing, video analysis, etc
4. Responsible for optical system design, simulation, calibration, or optical characteristic analysis, microscopic optical system imaging quality evaluation, super-resolution reconstruction, image enhancement, computational imaging algorithms, and software development in the fields of industrial micrometer/submicron defect detection, critical dimension measurement, and optical processing
[embodied intelligence]
1. Responsible for the research and development of embodied world reconstruction
2. Responsible for the research and development of multimodal data generation
3. Responsible for the development of an embodied autonomous training platform for generalizable robot tasks (such as grasping, manipulation, and movement skills)
4. Responsible for the development of a large model for multimodal embodied task planning
5. Responsible for the development of a embodied agent framework based on large models
【 3D Large Model 】
1. Responsible for the requirement analysis and algorithm implementation of Huawei Cloud for 3D reconstruction and 3D generation scenarios, including but not limited to AIGC related algorithm design for 3D content reconstruction, 3D positioning, 3D editing, and content generation. Design generation solutions that integrate multimodal information such as text, images, videos, and 3D, such as 3D reconstruction, 3D multimodal understanding, 3D generation, 4D generation, etc
2. Responsible for tracking and researching 3D understanding and generating cutting-edge issues and academic trends, exploring potential customer needs, and providing enterprise oriented technical solutions based on actual scenarios
3D Modeling and Scene Understanding
You will engage in large-scale 3D modeling/analysis/multimodal data pre training techniques, including but not limited to:
1. 3D modeling and full element semantic monolithic technology: Based on various data sources such as aerial images, satellite images, street view images, point clouds, and GPS/IMU that integrate air, space, and earth, it automatically integrates and reconstructs high-precision 3D models (including textures, materials, etc.) indoors and outdoors, and forms a fine-grained semantic monolithic model of all elements, achieving high realism, measurable light and shadow, and editable requirements
2. Change detection and automatic local update technology for 3D models: By using historical patterns and current data, automatic detection of model changes (geometric or texture changes) is achieved, and rapid local updates of the model are automatically performed
3. AI repair of 3D models: By using AI to automatically detect model defects such as broken surfaces and scratches, and utilizing techniques such as programmatic generation, template replacement, and geometric texture repair, the model is automatically repaired. The similarity between the repaired model and the real scene is high
4. Lightweight and visual rendering of 3D models: Based on techniques such as patch simplification, texture baking, component merging, and LOD partitioning, the model is lightweight without affecting the visualization effect. A new generation of rendering engine technology for end cloud fusion is also studied to achieve real-time and efficient visualization rendering of city level 3D models across platforms and multiple endpoints
5. Based on a variety of data sources such as aerial images, satellite images, street view images, point clouds, mesh, and GPS/IMU that integrate space, ground, and air, we automate the high-precision 3D scene understanding of large scenes (buildings, roads, trees, water systems, etc.)
6. Based on the integration of aerial images, satellite images, street view images, point clouds, mesh, GPS/IMU and other data sources, realize the training of multi-source data pre training models
7. Based on the integration of aerial images, satellite images, street view images, point clouds, mesh, GPS/IMU and other data sources, the alignment of NLP models and multi-source data pre training models is achieved
Autonomous Driving Fusion Perception Algorithm Engineer
You will be engaged in the design and development of autonomous driving perception algorithms for commercial vehicles, including but not limited to:
1. Responsible for the research and development of fusion perception algorithms for commercial vehicles (including but not limited to ports, mines, and other scenarios)
2. Responsible for the implementation and iteration of autonomous driving perception algorithms in actual products, and building Huawei's competitiveness in the field of autonomous driving for commercial vehicles
3. The use of mainstream autonomous driving sensors includes laser radar, millimeter wave radar, cameras, etc
4. Perception tasks include but are not limited to: object detection, multimodal segmentation, tracking, temporal prediction, multimodal fusion, etc
Deadline: December 26, 2025