Manuscript source: Alibaba Cloud Developer Community
In the spring of 2017 , 40 -year-old He Changhua made a decision: he returned to China to take a risk, give up the " Jiangshan " that he spent 12 years in Silicon Valley , and went to Hangzhou, thousands of miles away, to become the chief architect of Ant Financial Computing Storage. "It used to be that the motherland needs me, but now it is more in fact that I need the motherland. "
In an interview with CCTV in December 2017 , he said so. As a young top engineer in the Chinese community in Silicon Valley, He Changhua is a Ph.D. from Stanford University who entered Google from Cisco in 2008. As one of the core technology leaders, he has developed a new generation of Google's " caffeine " search engine with the team in 7 years . Won the company's highest technology award. But if he returns to China, he will be faced with a career " temptation " that the " tech men " cannot resist and Silicon Valley cannot give - hundreds of millions of users frequently use the scenes and products that they participate in the design every day, which almost represents infinity The potential and endless challenges. " Everyone wants to try their limit. I think it's here to really try whether they can achieve a limit. " He Changhua said.
2017 year-end, the ants gold dress at ATEC Congress for the first time published the technology of the future layout - "the BASIC" strategy , namely Blockchain ( block chain ) , Artificial Intelligence (AI), Security (security risk control), IoT (matter In the five major areas of networking) and Computing , AI artificial intelligence occupies a place in the main push direction. After returning to China, He Changhua led the team to develop and launched the ant real-time intelligent decision-making system; built China's first financial-grade distributed graph database Geabase ; developed a new generation of data technology architecture and financial-grade core computing engines. The pace after returning to China is obviously much faster than that of Silicon Valley, but He Changhua said that he is quite satisfied with the state of the whole work. The key is that what we are doing now is very breakthrough: "In the past, we often joked that C2C is Copy to China , which is done abroad. We will copy their answers again, but what we are doing now is something that everyone is exploring and leading the way. "
One of the leading things He Changhua said is the open source machine learning tool SQLFlow - combining difficult AI with simple SQL , which greatly simplifies the threshold for data engineers to use AI technology. It was the AI Infra team led by He Changhua that developed SQLFlow . SQLFlow allows users to describe the data flow and AI structure behind the entire application or product with only a few lines of SQL code , greatly reducing the threshold for using artificial intelligence. " Our purpose of doing this is to hope that all business personnel can directly use machine learning in the future. This is also in line with the democratization trend of domestic machine learning, which is to let all people be able to use machine learning. " He Changhua Say. 2019 Nian 5 Yue 6 days, ants gold dress deputy CTO Hu Xi announced: " the next three years, AI capabilities will become the basic capability of every skill we hope that through open source. SQLFlow , allows technicians to call AI likeSQL is as simple as that. " After the open source SQLFlow not only been used in marketing Alipay recommended scene, will soon be extended to similar " bit trip " such other industries.
If we jump out and look at it, the bright spot of SQLFlow is actually a more ambitious and ambitious plan - the fruit of an all-real-time big data intelligence platform . With the increasingly complex integration scenarios and massive amounts of data in the financial industry, the need for a new generation of data technology architecture and financial computing core engines has become more and more urgent. " Just like the real-time intelligent decision-making platform, the real-time calculation delay of the next-generation financial computing core engine should be low enough, the calculation speed should be fast enough, and the data should flow in real-time feedback results. " He Changhua said.
At Ant Financial, the latest task of He Changhua's team is to develop a new generation of computing engine and build a " completely real-time " big data processing system. This platform is a difficult " no man's land " . It is an area where many top research teams in the world want to do it, but no one has come up with a forming plan. Due to the diversity and complexity of offline life scenarios, this is a more challenging task than building the Google real-time search engine " caffeine " .
This task is considered to be the cornerstone of future technology. This " high-efficiency big data chassis " that brings together data, calculations, and intelligence is called Big Data Base , and its Chinese name is "the cornerstone " . He Changhua’s prediction is: “In the next three years, most computing scenarios will involve artificial intelligence, that is to say, in all aspects involving ant financial services, artificial intelligence will be used in 80% of the places. ” Many scenarios are compelling. Next, the development of artificial intelligence technology in the past decade is actually due to the huge amount of data and the increase in computing power. Whether a technology can achieve a qualitative lead depends on whether the underlying layer has sufficient computing power and fast enough model iteration. He specifically mentioned the new natural language processing model released by Google a few months ago. " The super computing power of this model is far ahead of similar companies in natural language processing . "
In He Changhua's view, the massive user base and data, the combination of AI and finance, will be the " episode scene " of the rise of China in the new era : " If you can participate in this scene and make some contributions, it will be beneficial to your life. A great improvement. " When a highly data-based society arrives, our ability to process and apply massive amounts of data directly determines whether we can do more based on data, catalyze higher intelligence, and push the human society downward. One-stage development.
He admitted that he is still a little " universal feelings " One of the most admired person is Andy Herron · Musk: " I think Musk is a legend, he did the things that many people can not imagine, dare to try, But things that are beneficial to the progress of mankind as a whole. " He Changhua now has to face hundreds of millions of users in his daily work. From inclusive finance to serving small and micro enterprises, he has a sense of responsibility that benefits the people: " I think people who engage in technology will ultimately this dream is - through technology changed the world. "
" When you think this matter is meaningful, it's worth it to be a little bit tired. " He Changhua, who rushed to the interview site after a long flight, looked a little tired because of a cold and staying up late, but he whispered about it. When making a vision, you can clearly see the light flashing in his eyes — that is the goal and passion of a scientist, and the eyes gazing into the future.