Huawei's "genius boy" top salary value! Realize the first large-scale commercialization of AutoML in the world.

From 2065438 to June 2009, Huawei launched the "Genius Boy" project, which is an important part of Huawei's founder's strategy of "dragging the world forward".

Since the project started two years ago, a total of 17 people have been selected, and key achievements have been made in some positions.

In an article in Huawei's internal magazine Huawei People, the achievements of the first batch of "gifted teenagers" Zhong Zhao were mentioned. In less than a year, Zhong Zhao and his team applied AutoML technology to tens of millions of Huawei mobile phones, realizing the first breakthrough of large-scale commercialization of AutoML in the industry. In the second year, the end-to-end pixel-level AutoML pipeline will be developed, which successfully reduces the complexity of video photography prototype algorithm by a hundred times and breaks through the limitations of industry and academia again.

In the self-report, Zhong Zhao said that basic research and commercial landing were carried out at the same time, and the latest technology of AutoML was used in commercial operations, which can be said to be making weapons while fighting. "Maybe only when it really goes down to the roots can this seed withstand the baptism of wind and rain and break through the ground."

AutoML's first large-scale commercialization in the industry

As a young team leading Huawei to explore in "no man's land", talented teenagers are regarded as stormtroopers who conquer world problems internally.

According to the email previously released by Huawei, the salary of gifted teenagers is paid according to the annual salary system, which has three grades, namely, 896,000-10.008 million yuan,14.05 million yuan-15.65 million yuan and18.2 million yuan-2,065.44.

Although the attractive salary is offered, the recruitment standard of Huawei's "gifted boy" is very strict, which generally requires about seven rounds of processes such as resume screening, written test, initial interview, supervisor interview, ministerial interview, president interview and HR interview. If there is a problem or poor performance in any link, you may lose the opportunity to enter Huawei.

Zhong Zhao is the employee who won the highest position in the "gifted boy" project.

According to public information, Zhong Zhao was born in 199 1 and studied software engineering at Huazhong University of Science and Technology. In my junior year, I won the first prize in Hubei in the 20 12 National Mathematical Model Competition for College Students. Then I went to the Institute of Automation of the Chinese Academy of Sciences to study for my master's degree and doctor's degree, both of which are "pattern recognition and intelligent systems".

In the readme, Zhong Zhao mentioned that influenced by his father, he was exposed to basic programming in kindergarten and primary school.

"My father was a group of students in the early days of Peking University. Before that, he did research in Institute of High Energy Physics (Institute of Atomic Energy) of Chinese Academy of Sciences. He is a student of Qian Sanqiang and He Huize. His earliest research on nuclear physics was to make two bombs and one satellite, mainly hydrogen bombs. They needed computers to do calculations related to nuclear physics, and later they turned to the computer field. There are some things in my father's works that I can touch. I still remember looking for him in the computer room and office, and I had to wear a shoe cover to avoid static electricity. After my father got off work, I played with the backup computer (LOGO language) and searched for new games secretly installed by college students in the DOS system. So my interest was cultivated in my father's subtle influence. "

In 20 18, Zhong Zhao went to the United States to attend the CVPR(IEEE International Conference on Computer Vision and Pattern Recognition), and happened to find that Huawei was deploying AutoML (Automatic Machine Learning) technology. Strong computing power, platform support and real business scenarios attracted Zhong Zhao to join Huawei.

After joining the company, Zhong Zhao's first task is to study how to make up for the shortage of optics by using algorithms, so as to achieve the effect that mobile phone photography surpasses SLR photography, that is, through AutoML technology, under the premise of ensuring the effect of taking pictures and drawing pictures, the algorithm is simplified to meet the requirements of product power consumption, speed and other indicators.

"Under the current circumstances, the hardware constraints we face are very harsh, but through two months of hard work, we have achieved remarkable results in solving problems." In Zhong Zhao's self-report, from 20 19 to 202 1, Huawei's camera algorithm made a breakthrough on M and P series mobile phones, in which AutoML system or algorithm played an irreplaceable role.

At present, AutoML technology has become the core competence of the department, and also supports the key business of many media such as video, ARVR and river map.

Activate the team to solve "world problems"

Zhong Zhao's "genius plan" is just a microcosm of Huawei's strengthening of R&D layout.

"These teenagers are like' loaches', drilling into our organization and activating our team." Ren once said that in the next 3-5 years, I believe that Huawei will be brand-new, and all of them will be "guns for guns", so we must win this "war".

Huawei has publicly stated in a video how to cultivate "talented teenagers".

Tian Qi, chief scientist in Huawei's cloud artificial intelligence field, said in the video that "talented teenagers" mostly refer to doctoral students who just graduated from the age of 25 to 30. At this time, their physical strength, intelligence and innovation ability are the best.

In the process of training, Tian Qi mentioned that the first step is to communicate and understand your own strengths; The second step, "Good steel is used in the blade", introduces the pain points and difficulties of the business to them, and completes the innovation link from 0 to 1 to N; Third, encourage them to find and solve problems on their own initiative; Fourth, provide a good laboratory atmosphere and encourage free thinking and discussion. Huawei's expectation for talented teenagers is to "do the strongest contact research, put the results of these basic research into the industry, and finally precipitate on the AI platform".

Huawei official website shows that in addition to Zhong Zhao, Zuo Pengfei and Li Yi, who were selected as Huawei's "Genius Boys" in 20 19, have been engaged in cloud storage research and formal verification of operating systems in Huawei.

In addition to talented teenagers, Huawei has invested in training scientists and other talents for many years.

According to public data, Huawei has at least 700 mathematicians, 800 physicists, 120 chemists,15,000 personnel engaged in basic research and more than 60,000 product research and development personnel. At the same time, Huawei has also cooperated with more than 300 universities and 900 research institutions and companies around the world, implemented 7,840 projects, invested 65.438+0.8 billion US dollars, and signed more than 654.38+0.000 paid R&D cooperation contracts.

When asked how to evaluate the effectiveness of basic research and what kind of scientists he has in mind, Ren once said, "Basic research cannot be evaluated by quantitative assessment methods. We should look at what scientists say from a long time axis, and we can't care whether all the contents have practical significance. Fermat's last theorem was proved in 350 years, and it didn't make much contribution to the French economy. If we look at the problem so narrowly, how can human beings explore progress? But we have a constraint that the direction should be roughly on the main channel of the company's progress. We should be more tolerant of scientists, and we will be more tolerant of things we don't understand as long as they are roughly aimed at the main channel. "

"We should strengthen the investment in basic research, and hope that the cost of basic research will be allocated more, such as 20%-30%, from the total R&D expenditure of $ kloc-0/50-20 billion per year, so that about $3-4 billion will be invested in basic research every year." Ren said.