Scientists who left BAT, was snapped by VC

Observing the bodies of these scientists in BAT, almost clear, have a computer professional background and abroad experience, and the professor or researcher of the famous school is standard, academic attainments, or technically excellent skills. These scientists have the ability or potential to convert scientific research results into commercial kinetility, so the Internet’s large factory does not rush to compete for talents. For example, the vice president of Ants, AI chief scientist is in front of the ants, is the deputy professor of the University of Pudi University, Ali In order to dig, the paint is far from 5-10 times the original salary. In addition to the rich salary, the resources given by the Internet Factory are also a temptation that cannot be resistant: byte beating is specifically established in the AI ??laboratory and the artificial intelligence laboratory, Baidu Research Institute’s cognitive computing laboratory, Silicon Valley artificial intelligence A series of technical platforms such as laboratory, large data laboratory, robot and automatic driving laboratory, which gives the stage that "the world" scientists play. Alibaba Humper has begun to build a leading technology research center in all over the world, covering quantum computing, machine learning, and machinery intelligence,, financial technology and other industries. Tencent Ai Lab computer vision, speech recognition, natural language processing, machine learning also gives Make aieverywhere this vision is very imagined. It is not denying high salary and rich resources. These are the reasons for excavation of these scientists. More importantly, these scientists have high autonomy in the Internet, can accelerate their research results. For the Internet major factory, scientists’ joys can support the strength of the company and speed up the product iteration. Ma Weiti served as the vice president and the Director and the Director of the Organic Intelligence Laboratory, the AI ??strength of the Physical Intelligence Labs; the paint is returned to the country for the vice president of Alibaba, 2015 created and led the antiques of the artificial intelligence team; Wu Weida was Baidu in 2014 Appointment as a chief scientist, it is comprehensive responsible for Baidu Research Institute, especially Baidu Brain plans … The Internet is giving them a territory of their scientific research. These scientists are not expensive, and efforts have made technology to serve the industry more efficiently. Second, the BAT, was snapped by VC, although many "Internet players" regard BAT as the highest point of career, but for scientists in the top of many radiopes, BAT may be just a springboard in their lives, even a short-lived The experience, "get it, you can’t stay" becoming the normal state of BAT scientists. Children’s Day, six years ago, gain a master’s degree in physics in Chicago, and the Ph.D. in Standard’s Ph.D. leaving the California Mountain View, sitting on China’s flight. He has worked in IBM, Google, returned to China, waiting for him to be the position of "Alibaba Cloud’s Artificial Intelligent Scientist". After entering the job, Mili is responsible for incubation of various artificial intelligence applications, from there is no formal data incubator team, from the initial road signal light optimization system to the later ET city brain, ET industrial brain, ET agricultural brain, 闵 万 万 立"Khan Music". However, in four years, Yanyi has made a good relationship with Ali Cloud with a "mountain scene". Jia Jiaya has left Tencent, founded the application innovation of Science and Technology, focusing on deep learning and computer vision and other technology, and completed $ 200 million in B round finance in June this year. Another Ph.D., artificial intelligence expert, internationally renowned machine learning expert Yu Kai also waved sweat in Baidu, Yu Kai’s good at the field of speech recognition, computer vision, etc. The deputy dean of the hospital, served as senior director responsible for Baidu pictures search products, and the team he led three times won the "Baidu Most Award". After 3 years after Baidu, Yu Kai or the departure from Baidu. Not only that, Baidu Chief Scientist Wu Weida also left from Baidu, because the high-rise of the Baidu Research Institute is frequent, Baidu is also called "artificial intelligent Huangpu Military Academy". Tencent also can’t escape the destiny of "scientists." International well-known scholars in the field of machine learning and big data analysis, Zhang Jieguan, the State University of New Jersey, Joined Baidu Deep Studies Research Institute in 2013, and joined Tencent in March 2017, as AI LAB Director. But after Tencent two years later, Zhang Yun left again, and selected to return to the academic circle and continued academic research in the field of AI. Zhang Yong, Yu Wanli, Yu Kai, Zhu Sheng fire, Ren Xiaofeng … These BATs of the scientists digging back, but they have not been able to stay in the big factory. In BAT, due to the direction and demand of technology research and development at all phases of the company, the direction of the company’s talent retainer will be transferred, which is uncontrollability. At this time, there may be contradictory conflicts in the development direction of the scientist itself. When the conflict cannot be properly adjusted, the departure will become an inevitable choice. For scientists, whether it is the maximum value of their own value, or the maximization of wealth value, entrepreneurship is a best way. After all, their famous school graduated, adequate people, and it is not as good as a "advanced work". The fact is that most of the scientists from Bat have embarked on the road of entrepreneurship. Yu Kai has created a horizon robot after leaving Baidu. This company is an embedded artificial intelligent core technology and system-level solution provider. The main business is to install "brain" for terminal equipment such as automatic driving, smart cameras. As of 10 June 20, 2021, the skyline robot has experienced 13 rounds of financing, and in nearly half a year, the horizon robot has been investing 7 rounds, with a total of at least 1.2 billion US dollars, strong enhancement capacity. Wuanta, who is the same as Baidu "away", has also selected entrepreneurship. Wu Yida’s Woebot has announced $ 90 million B round in July this year. After the financing of this round, the total amount of Woebot Health Health reached 114 million US dollars. The financing party includes Jazz Venture Partners, Temasek, Belles, OWL Ventures, etc. After leaving Ali, Mi Wanli led a venture capital fund, focusing on the surrounding of traditional industries.With intelligent cloud technology and capital injection blessing "two in one" Combined energized. According to media reports, titanium, Min Wan founded the VC fund raising has been put in place, the first phase of about $ 800 million. BAT exodus of scientists from the creation of the company is highly favored capital "darling", the amount of financing is also very impressive. Third, the "wisdom of the brain," the next leg of the "Science and technology are primary productive forces" era, this group of cutting-edge technology to master the most high-profile scientists, they are these enterprises as "smart brain" sophisticated people go, it means resources, capital, talent will bring together where. Careful inventory of scientists who flow away from BAT, in addition to their own businesses bend into the Council, is to return to academia, with his years of academic heritage to return to academia, watering a new generation of scientific and technological personnel, one personally promote research change. After beating Ma Weiying, vice president resigned byte position, followed his leadership in the old Microsoft Research Asia, Ya-Qin Zhang, adding intelligent Industrial Research Institute of Tsinghua University, engaged in personnel training and scientific research related work; paint far away from Alibaba after the entry Fudan University, served as Fudan Dean of innovation and industrial Research Institute of artificial Intelligence University; former executive vice president of Microsoft’s global Shum also last year as a part-time professor of Tsinghua University joined the Institute for Advanced, and for computer vision, computer graphics recruit doctoral students. Moment, artificial intelligence has been to promote industrial upgrading, the overall productivity jumped drive force as early as 2017, Kai-fu Lee said Innovation Works CEO on the future of artificial intelligence will replace 50% of working people. The "2020 China Artificial Intelligence industry White Paper" also shows that China within five years of artificial intelligence market size is expected to more than 400 billion yuan. And "new consumer" different "cultural" industries, AI industry is a technology-driven industry, this industry is not an ordinary entrepreneur roots in 2035 is not getting started, but the fate of entrepreneurial scientists can play a leading role in the integration and now all applications AI industry will need technical talent, a great gap in the market. It can be said, it is now scientific entrepreneurship is the best time ever, do artificial intelligence, big data this group of people, are also eagerly longing come from technology to specific application scenarios. From the industry perspective, the, AI core business is the AI ??scientists. But the domestic business environment is still conservative think research needs quiet, and entrepreneurs need to move. Liu also said that "scientific ideas and the company’s operations are very different, some scientists are not good at managing." China recently Jingwei Zhang Ying, a founding managing partner of "research-based recommendations to the nine founders of" scraper, text middle the problems scientists face of entrepreneurship: scientific or technical background in the field of their professional authority, profound technology accumulation, but also often because out of business relatively late age compared to the mobile Internet entrepreneur, founder of relatively larger; also because of science, strong technology-oriented, character is relatively easy to cure, easy-line thinking, more difficult to accept new matter; scientists tend to be forward thinking, like to think from a technical line of products form, but the market needs is often reverse thinking, according to business needs to go retrogression technical match. BAT exodus from scientists frequently get the favor of capital, the fundamental reason is that this is related to a "cost" of multiple choice questions. VC decide whether to invest in a company, will predict the value of this start-up companies, in this dimension, the most important thing is whether a company has in line with future trends of technology and products, although AI technology landed long, but wins in the foreground wide . Scientists from the Internet giant exodus, not only the real deal of technology, knowledge and experience more commercial aspects of both science and technology talent at ease, but also entrepreneurs know how flexible mind, there is an unusual entrepreneur not available contacts resources, these are their most valuable point. In recent years, AI has long been as good as 2015 venture so hot, in the case freeze the market environment, each began to tighten the purse strings, also shot more cautious. In comparison, those who voted through the "money someone brains" of BAT scientists combat, obviously have a greater chance of winning, but also more cost-effective choice. Technology determines the length of life of the enterprise business, the Internet giant behind the competition for scientists, in fact, is fighting to build technical barriers weights, and these manufacturers are attracted to scientists in recent years frequently away from home, perhaps the business and personal career pursue not fit, maybe not well compatible with academic and commercial, but for whatever reason, can not deny the fact: they always stand in the forefront of the industry, it is a group of people who touch the future.

Vite’s project, the method of packaging using Rollup package

Build a production version – library mode

Detailed setting

Vue-cli comes with WebPack’s packaging method, the played bag is a bit big, and Vite comes with Rollup packaging, this way is very small, there are some instructions for the official website, so after learning, It is very convenient.

Vite’s library project can be divided into two categories: (I am divided into it)

  • One is a pure JS project without html;
  • One is a project that can bring HTML (template), such as a UI library.

The following introduces the preparation and packaging methods, in fact, in fact.

Use Vite to build a project, here is a simple example:

// main.js

The code is relatively simple, just a demo.

If you want to pack, you can only have an export file, so the interior code structure is set.

Pure JS is good, the export output is good, then what should I do with template? In fact, it is also the same.

Create a project with Vite to create a test file:

// t-text.vue

  • Template section:
  • Code part

This is a library based on the UI library, using two third-party plug-ins:

  • One is Element-Plus, which is loaded on demand;
  • One is the NF-UI-Controller library they do.

Then we set an entry file main.js

If there are many components, we must consider the structure, here is just an example.

After the code is written, you need to set up vite.config.js

  • Entry: Specifies the entry file to be packaged.
  • Name: Name of the package
  • FileName: The name of the package file, the default is UMD and ES two files.
  • SourceMap: Do you generate .map files, default is not generated, if you need it, you need to set it to true.
  • RollupOptions: If the project references a third-party plugin, you need to set it here. If you don’t set it, the source code of the third-party plugin will be packaged, so that the package file is getting bigger. The third party’s plug-in will exist alone.

Vite will pack it according to the settings here, and pack them into two modes:

  • UMD: After packaging, the code is very compact, small size, but it is not easy to read;
  • ES: The packaged code and the code we wrote are very like, easy to read, but the volume is large.

If the project uses a third-party plugin, you need to do settings in External:

After setting, the code of the third party plug-in will be referenced in the way in import.
If you don’t set it, you will use the code that is used in the third party plugin, and take it out as the internal code to be packaged together, so the volume of the package becomes bigger.

Setting package.json is mainly published as a description of the resource package, you need to set the properties according to NPM requirements:

If you do not want to install a third-party plug-in when installing the resource package, you need to remove DependenCies and DevDependenCIs, otherwise the third-party plug-in that is automatically installed.

After setting up, we can pack it with Yarn Build.

The packaged file is this style:


Brief introduction to the steps:

  • First to website registration account;
  • To register an email to activate the account;
  • Open the terminal and log in with npm login;
  • Published using NPM Publish;

The space is limited, and the details will not be introduced.

I like to use Yarn to install resource packages because the speed is faster.

Package.json Set the name of the installation is the name of the installation, so you have a good name and you can’t repeat with the existing name.

Will there be such a need?
When writing, the library project needs a development environment, while writing code, while running.
After writing, you need to pack it by library project.
Then there is a test environment, or demonstrate the environment?

If you build a project write test and demonstration, then it is a bit trouble, if you can get more in a project.
At this time, you need to set different Vite.config.js.

The way to use the comment before the change is more troublesome. Now found that Vite provides "Mode" way, allowing us to do different settings in Vite.config.js.

According to the official website, we can build multiple .env. * Files to store different "modes":

.ENV: Development environment (default)

.env.project: Test, demo

.env.lib: library package

After setting it, we will modify vite.config.js

In Vite.config.js, first define different defineconfig spare, then use the loadenv read mode value, return the corresponding defineconfig according to the mode.

Note: You need to set a folder of different outputs, otherwise it will be overwritten.

Set the executed command, add the desired mode.

This allows you to completely separate different environments. If there are other needs, you can also add more modes.

Changeable package command

In this way, all of them do not interfere, nor do you have any writes.

After the project package is released to, you can use to access, XXX indicates the name of the resource package, such as: