Salary: 45-75K
Experience requirements: Experience is not limited
Educational requirements: Doctor and above
Job Description:
Job responsibilities
1. Participate in the research and development of the perception and decision-making algorithm of the robot system, and improve the accuracy and robustness of the embodied intelligent algorithm based on the self-developed robot platform;
2. Design scenarios and indicators to test the robot's embodied intelligence algorithm qualitatively and quantitatively;
3. Design a self-supervised/weakly supervised/active learning framework to improve the data utilization and algorithm iteration efficiency of the self-developed data closed-loop platform.
Job requirements
1. Doctoral degree, computer, software engineering, artificial intelligence and other majors are preferred;
2. Familiar with the principle of deep learning, understand the network structure, training paradigm and test indicators of common neural networks;
3. Understand common deep learning frameworks (such as PyTorch, Jax, etc.), and have project experience in network training and tuning;
4. Understand the concept of representation learning and have a certain understanding of self-supervised/weakly supervised learning paradigm;
5. Understand the cutting-edge algorithms in the field of computer vision (such as Segment Anything, Grounding DINO, Diffusion Policy, Gaussian Splatting, etc.);
6. Familiar with 3D computer vision, understand common models in 3D Vision field (such as PointNet, SparseConv, etc.);
7. Proficient in Python and C + + programming languages, with the ability to independently locate and solve algorithm and software problems;
8. Pay attention to the industry frontier and have a strong interest in the development of embodied intelligence;
9. Have a certain English foundation and be able to read English technical documents;
10. Proficient in using Git to develop collaboration tools.
Bonus points
1. Published articles in Top international journals/conferences in the field of machine learning;
2. Understand the CUDA parallel computing framework, and have project experience in using CUDA to implement custom operators;
3. Understand the principles of model tailoring and quantification, and have project experience in edge deployment.
Welfare policy:
1. Perfect training system, tailor-made training courses for employees;
2. Purchase five insurances and one housing fund according to national regulations;
3. Enjoy statutory leave, marriage leave, maternity leave and other paid leave according to regulations;
4. Enjoy annual leave calculated according to the length of social service;
5, rich and colorful staff activities, from time to time to organize staff travel, enrich staff life.
Address: No.7, No.3, Jiucun West Road, Jiushuikeng Village, Panyu District, Guangzhou
Working hours: 8 hours, two days off.