laitimes

News said that He Changhua, a P10 expert at Ant Group, left his job and returned home from Silicon Valley with a salary cut 5 years ago

News said that He Changhua, a P10 expert at Ant Group, left his job and returned home from Silicon Valley with a salary cut 5 years ago

Recently, it is reported that He Changhua, chief architect of computing storage of Ant Group, has left his job, but his whereabouts after leaving are temporarily unknown. The "China Times" reporter asked the Ant Group for verification, but the Ant Group officially did not respond to the matter.

Before joining Ant Group, He Changhua worked for Many years in Silicon Valley technology companies such as Google and Airbnb, and his resume was quite bright. He Changhua joined Ant in 2017 under the name "Xiao He". He Changhua has developed a real-time intelligent decision-making system, a financial-grade distributed graph database, and a financial-grade core computing engine led by Ant.

Silicon Valley veterans return home

Public information shows that He Changhua is a native of Hubei, studied at Tsinghua University, and later received a master's degree in philosophy from the University of Hong Kong and a master's degree in science and a doctorate in philosophy from Stanford University in the United States.

Before joining Ant, He Changhua worked in Silicon Valley for more than a decade, working for Cisco, Google, airbnb (Airbnb), and his career resume is very eye-catching.

He Changhua worked at Google for 7 years, as one of the core technology leaders, worked with the team to develop Google's new generation of "caffeine" search engine, won the company's highest technical award, and served as Google's senior chief software engineer. He changhua then worked at Airbnb for 2 years, responsible for the application architecture of the back-end system.

He Changhua has said that in fact, he has long been willing to return to China, but because his work belongs to the "infrastructure" part, only companies with strong technical needs will need it at a certain stage of development.

During Airbnb's work, Hu Xi, then vice president of Ant Financial, approached He Changhua, and after a few conversations, He Changhua decided to return to China to join Ant.

In an interview with CCTV News, He Changhua said that one of the reasons hu Xi gave him the most was that "the platform ant really promotes the lives of ordinary people every day", and although Airbnb is also a popular technology company, its products are far from so many users and such a high frequency of use.

In May 2017, He Changhua officially joined Ant, that year He Changhua was 40 years old, and his salary after joining Ant was much lower than in the United States. At that time, the scale of ants was more than 8,000, and the proportion of employees with overseas study backgrounds accounted for 8%.

Lead multiple innovations

After joining Ant, He Changhua led the team to develop and launch the Ant Real-time Intelligent Decision-Making System, which can make real-time judgments on financial fraud and money laundering. He Changhua's team also used graph computing and graph database technology to build China's first financial-grade distributed graph database Geabase, and developed a new generation of data technology architecture and a financial-grade core computing engine.

An important breakthrough in the new generation of computing engines developed by He Changhua is SQLFlow.

The AI Infra team led by He Changhua developed sqlflow, an open source machine learning tool. SQLFlow is easy to learn and easy to use, simplifying the threshold for data engineers to use AI technology, thereby promoting more AI application scenarios to be discovered.

Previously, He Changhua participated in the development of a new generation of search engine "caffeine" in Google, which solved the delay problem of traditional search engines, and users could search for changes in web pages in real time.

He Changhua also wants to build a fully real-time big data processing system in Ant, which is a more challenging task than a real-time search engine and has high requirements for computing speed and ability.

MapReduce, a commonly used big data computing model before, has data processing that is "sharded", and this batch processing model will bring delay problems. To achieve true real-time, it is necessary to break the boundaries of this batch processing.

He Changhua believes that the "real-time system" is not only fast, but also means that online transaction processing and online analysis processing are integrated. For example, a transaction with Alipay requires instant query and addition and deletion of records, which is achieved by online transaction processing. The data analysis of user behavior characteristics is online analysis and processing. With the development of business scenario requirements, the timeliness requirements of online analysis and processing are also getting higher and higher.

The real-time feedback of online analysis and processing can analyze the user's characteristic data in a very short period of time to determine the risk, and is also an important technical means of financial risk control.

"Real-time systems" are also a fusion of intelligence and data systems. Many companies separate the digital warehouse from the machine learning platform, take data from the digital warehouse, and then put it on the machine learning platform to train the model. The drawbacks of this model are also gradually exposed. He Changhua pointed out that real-time data inflow, real-time training of models, real-time on-line decision-making and feedback of data, if this line can be fully opened, it will produce immeasurable value for the business.

According to Alibaba Group's e-commerce new media platform "Tianxia Netshang", more than 60% of Ant's nearly 10,000 employees in 2018 were engineers, and half of Ant Financial's 6 Alibaba partners at that time were from technical backgrounds. In recent years, ants have also attracted many returnees, and He Changhua is one of them. The future of these technical talents is worth paying attention to.

Read on