165438+1October 9-12, Beijing National Convention Center, the 6th TOP 100 Global Software Case Study Summit. In 4 days, I gained insight into the case practice of the technical director of 100.
20 17 TOP 100summit is still five special sessions in parallel, and 15 topic direction comprehensively shows the practices and solutions of all dimensions of software development life cycle.
Session 1: Experience Design/Product Innovation/Operation Drive
Selected cases
● "Balance creativity and technology to create innovative products"
Ruthia He ——Facbook product designer
Case value: The design process is like a journey to find a balance between product goals, technical realization, creativity and user experience. Dealing with tight resources is an art. For example, you need to remind yourself at all times what your product goals are, but the technical resources to achieve your product goals are probably insufficient. It may also be that the designer's inspiration gave birth to a unique idea, but this idea may not be accepted by everyone. This case will tell you how to find the "perfect balance point" of product design according to the lecturer's product design experience in Silicon Valley.
Design thinking behind "one-dimensional painting"
Chen Xiaochang —— Director of Design Center of Tencent User Research and Experience Design Department
Case value: On August 29th, the circle of friends was screened by the beautiful paintings of "children". In just half a day, 5.8 million people participated and raised more than150,000 yuan. The Internet is already changing our public welfare undertakings. Science and technology link trust, public welfare design, how to do it will be better. This case will restore the whole communication event, show you the design thinking behind the "one-yuan painting purchase", and discuss how to create more value for public welfare with service design in combination with the design team's experience in supporting Tencent Public Welfare 10.
● "Uncovering the era of artificial intelligence terminals-thinking and definition of Tmall Elf"
Yi Ru —— Head of Intelligent Terminal in Artificial Intelligence Laboratory
Case value: The emergence of Tmall Elf represents Alibaba's thinking and exploration of intelligent terminals in the era of artificial intelligence. Intelligent terminal not only improves the user experience, but also lowers the threshold of use, which means the closure and strengthening of the terminal ecology. The relatively closed ecology in the era of artificial intelligence means that manufacturers want to provide a quality service experience, which is difficult to complete without a terminal. In the future, cloud integration will become a new pattern of the general trend. This case considers how to use its own advantages to define and land terminal products.
● "Customer First-Stand Out from the War of Smart Home Terminals"
Chen Ya-Amazon Senior Engineer
Case value: As the entrance of smart home, the terminal of smart home is the main position for giants to seize the market. So what makes Amazon, an e-commerce company that started as a retailer, stand out in this war and occupy 70% of the terminal market with overwhelming advantages? This case will take the Echo product as an example, and analyze it from two aspects: product design and development management mode, how Amazon penetrates customer obedience into all stages of the product, and how to suppress Google, which is good at technology. At the same time, this case also makes an exploratory analysis of the current domestic smart home terminals.
● "Growth behind Didi's new business"
Li Sen-Head of Didi Growth
Case value: This case will describe the thinking and practice of the sharer's growth business from 0- 1, such as the innovation of car owners hitchhiking, the growth of users of bus products, the cold start of minibus products, and the business of express trains in Chongqing. Starting from the logic of growth, we will introduce some tried-and-tested growth graspers through resumption of business, and introduce the business of Didi Express and minibus from 0 to 650.
Special Session 2: Engineering Culture/Team Growth/Performance Evaluation
Selected cases
● "The Science Behind Art: Five Years' Course of Riot Games Data Team"
Li Renjie -riot games Data Director
Case value: Based on the five-year mental journey of Riot Games data team, this case introduces how to build a world-class big data team from scratch, how the team's work and vision grow and evolve every year, as well as gains and detours. Taking a case selected every year as a sideline, this paper introduces how League of Legends, the most popular game in the world, uses data to enhance the player experience, supports and helps the business decisions and operation plans of various departments of the company, and subverts traditional products with machine learning and artificial intelligence.
● "How to ●" Google uses OKR to help the team challenge the impossible task "
He-Google Product Manager "It only takes two days to change from a traditional project to an agile project".
Case value: As one of the top technology companies in the world, Google has challenged many high-tech tasks that seemed impossible before, such as AlphaGo, Google Translation, self-driving cars, Tensorflow, TPU and so on. How does Google organize and motivate teams? How to ensure the team Qi Xin to work together and sprint in the same direction? This case comes from the first-hand experience of Google's current product manager in the US headquarters. He will share the success of Google management from the aspects of mechanism, humanity, process, decision-making method, product policy and company organization.
● "The Evolution and Transformation of Lean Kanban of Huawei Hundred-member Team"
Chen Jun —— Huawei Agile Lean Expert
Case value: Facing the surge and rapid change of market demand, R&D team needs to be flexible and improve R&D efficiency with limited manpower. It is believed that the introduction of lean kanban can effectively help improve the efficiency of R&D. This case describes the evolution of lean kanban in Huawei's 100-member team, from establishing kanban (four practices) to operating kanban (four practices), and achieved a small victory. Then the team encountered difficulties, stagnated or even regressed, and re-examined the improvement with the team in the face of difficulties and embarked on the right path.
● "It only takes you two days to make an agile transition from a traditional project"
Gu Yue-Senior scrum Master of Ping An Technology
Case value: Agile transformation not only applies a new set of processes, but also changes people's way of thinking and working, and even changes the organizational structure of enterprises. Is there a shortcut to transformation? How did Ping An Technology's two-day Quick Start Workshop become a powerful wrench to switch from traditional track to agile track? The case will be revealed for you one by one.
● Atypical Agile: 10 day version.
Zuo Yangmei: Master scrum of Zhongxing
Case value: "Fast" is relative. The traditional telecom field still adheres to strict addition rules and security requirements, and follows the basic process of "demand-realization-release-upgrade". In this case, the concept of process deliverable is introduced from the perspective of reorganizing user value, which realizes the deep participation and rapid feedback of customers. Re-examine the core practices of typical agile processes, aiming at "quickly verifying customers' product assumptions", eliminate practices such as automated testing and continuous integration, introduce hand-drawn minimum real delivery, and introduce data simulation and slicing functions. To some extent, this case is an extended delivery project of Design Sprint in the field of telecommunications.
Session 3: Architecture Evolution/Engineering Practice/Big Front End
Selected cases
● The architecture practice behind the 6 18 promotion gateway carrying one billion calls.
Chief Architect of the Open Platform of JD.COM Shopping Center in JD.COM, Dong Wang
Case value: The open platform of JD.COM Mall is promoted on June-October-August every year, which ensures the massive call of nearly a thousand different service interfaces, and at the same time ensures that the service interfaces do not interfere with each other and can quickly respond to any complicated situation. Stability and quickness are the goals we have been pursuing. This case will share some common methods in practice, such as isolation technology, caching technology, SQL optimization, degradation, current limiting and so on. What about JD? COM team applied these technologies to every preparation, ensuring the smooth passage of 6 18 every year.
● The structural transformation of the new generation trading system of Shenzhen Stock Exchange
Yu Huali —— Chief Engineer of Shenzhen Stock Exchange
Case value: The business system at the core of the industry has strict requirements for continuous and smooth operation. How to upgrade these core business systems to meet the needs of business development and technological progress is a difficult problem faced by many CIOs and their R&D teams. This case shared the successful experience of Shenzhen Stock Exchange in the core system, especially the high availability and high performance real-time processing system, implementing structural transformation and smooth upgrade of IOE, open platform and open source technology, distributed processing, high availability and low latency design, and shared how to ensure a safe and smooth upgrade in this comprehensive restructuring structural transformation and lead the smooth upgrade of the whole market.
● "Hungry?" The overall service is the transformation in different places.
Is Li Shuangtao hungry? Chief architect of middleware team and multi-activity project in different places.
Case value: This case describes how to coordinate the work of business team and middleware team from design to official launch, and how to transform the whole business safely and smoothly, so that the business can be transformed from single-room service to multi-room multi-activity service. When the computer room level fault occurs, the service provider can route users to a healthy computer room, thus ensuring the normal operation of the business when the fault occurs and reducing the huge losses caused by the computer room level fault.
● "Uber is a business, micro-service innovation practices medical transformation from 0 to 1 digital.
Shi Xiaoyu-Head of Uber University of Technology
Case value: This case will share how to realize a highly available system from 0 to 1 and solve the practical Uber of enterprise business problems. Through the specific project requirements and system architecture, including payment system, billing system and policy system, this paper analyzes how to complete these systems end to end. How to complete the process from 0 to 1 in just two years has become a very important performance growth point for Uber. At the same time, it has grown from a team of 6 engineers to nearly 40 people.
● "Exploration of Xiaomi Direct Service Platform and Future Form of Mobile Service"
Dong —— Head of Xiaomi MIUI System Framework Team
Case value: At present, there are some shortcomings in the bearing forms of mobile services, both applications and web pages, which bring inconvenience to users and have a certain impact on developers themselves. How to distribute and use services more efficiently is a topic of great concern to the industry. Xiaomi has also made some explorations in this respect, and launched technical platforms such as direct service, aiming at solving some problems existing in the case of traditional applications and web hosting services and improving the efficiency of users and developers. This case mainly focuses on Xiaomi's direct service platform, and talks about Xiaomi's ideas and some practical achievements at present.
Special Session 4- Data Science/Artificial Intelligence/Data Driven
Selected cases
● "How NFCU Bank of the United States uses big data AI to open the road to transformation"
Jiang Xiaodong-NFCU Financial Data Architect
Case value: Value Bank of America is a Fortune 200 company. By the end of 20 16, it had 280 branches around the world, with assets exceeding 740 billion US dollars, more than 60,000 members (customers) in the United States and14,000 employees worldwide. How to manage the daily cash flow of branches and ATMs under such a huge global volume, and integrate the cash storage, transfer and withdrawal between headquarters and branches, branch counters and customers, and customers and ATMs, determines the results and efficiency of settlement and cash flow supervision and management between banks and cash trucks, central banks and banks. This case opens up a very meaningful case for large traditional financial enterprises to implement big data and AI projects, and will share the ways and means of NFCU Bank to solve corporate cash flow management by using big data and artificial intelligence algorithms.
● "In the era of artificial intelligence, how did the intelligent recommendation system of second-hand trading platforms evolve?"
Sun Xuan, head of architecture algorithm department
Case value: The referral system was established from 0, and gradually developed according to different stages of business. In the development process, it has gone through the stages of global non-personalized recommendation, personalized offline recommendation, personalized real-time recommendation, machine learning ranking recommendation and so on. This case meeting will explain in detail the reasons of different development stages and the evolution of the architecture, so that the audience can have a deep understanding of the intelligent recommendation system of the second-hand trading platform.
● "Prophet: Artificial Intelligence Helps Fintech Anti-fraud, Making Black Products Nowhere to Hide-How Big Data and Artificial Intelligence Help the Risk Control and Defense System"
Wang Ting —— Pleasant Loan Data Scientist
Case value: Prophet is an anti-fraud cloud platform based on pleasant loan, and it is an anti-fraud solution for the entire Fintech industry. With its powerful financial data capabilities, anti-fraud intelligence and online customer service capabilities, Fintech enterprises can solve a series of problems such as credit application fraud, financial intermediary identification, gang monitoring/early warning, and provide financial technology enterprises with stronger credit evaluation, risk control and accurate customer acquisition. This case will share how to use artificial intelligence to achieve the above functions during the construction of anti-fraud cloud platform.
● Some applications of machine learning and statistical modeling in online-to-offline scenarios.
Zhang Jian —— Head of 3M Data Technology
Case value: Online to offline is an important trend in the future. Data mining and machine learning have been widely and maturely applied in online software development, recommendation matching, user analysis and other fields. However, offline and online data fusion and optimization have just begun. This sharing will start from the specific case of online and offline retail, build an online and offline data feedback and optimization system, and apply statistics and machine learning methods such as A/B testing, in-depth personality recommendation and reinforcement learning to achieve a series of specific goals such as improving data analysis efficiency, understanding user behavior and increasing offline income.
● "Lenovo Big Data Helps Lenovo's Business Transformation and Upgrade"
Yu Chen Tao —— Senior Director and Principal Investigator of Big Data Division of Lenovo Group
Case value: The fourth industrial revolution driven by digital transformation has begun, which has opened a new route of big data, cloud services and intelligent technologies in parallel. When enterprises win opportunities, they also face many problems: the data of various systems within enterprises cannot be shared, and the phenomenon of data blocking is serious, which directly leads to the inefficiency of procurement, production, logistics and sales. This case shared how Lenovo solved the above problems with the help of big data, industrial Internet 4.0 and Made in China 2025 under the premise of controllable cost, and achieved rapid development with the help of the tuyere.
Special Session 5- Quality Management/Intelligent Operation and Maintenance /DevOps Special Session
Selected cases
● "How can unmanned testing help JD.COM improve the efficiency and quality of product testing?"
Jin Yang, head of B2B product quality team in JD.COM.
Case value: With the development of business, the system usually goes through the process of individualization, service and platformization. In the long-distance evolution of the system, whether it is small demand or big change, every time it goes online, it is accompanied by a lot of regression work. Even experienced test drivers are not 100% sure that there will be no problem. In the Internet industry with short iteration period and high release frequency, we have been trying to explore and practice how to ensure product quality and improve user experience in frequent online launch. This case describes an efficient regression testing method and its practice in improving the efficiency and quality of product testing.
● "Ali Mobile DevOps Practice"
Lu Yiyuan, head of Alibaba platform products
Case value: the mobile development model has entered two levels of differentiation: the R&D model of super-large APP is project-oriented, and there are many people and modules in R&D collaboration, which need to build, test, publish, operate and maintain a complete DevOps system; Some innovative and experimental apps are more suitable for testing and verifying business ideas in a faster way when the business model and business form have not been fully determined, so it is imperative to quickly build an app at the lowest cost. This case will share how Ali mobile technology has precipitated and solved these problems in the past few years.
● "Take Kafka as an example to explore the optimization method of large-scale stateful cluster"
Qin Lingying Employee Software Engineer
Case value: Dynamic load balancing and self-management of distributed systems have always been a difficult problem to solve. Most solutions are to migrate the whole application process and realize the load balance of hardware resources. This method is suitable for stateless applications, but it is not very effective for stateful clusters such as Kafka. Because migrating an application means migrating a large number of states, it is a long and expensive process. LinkedIn developed cruise control to solve this problem. Its main feature is that it can transfer some states according to the characteristics of the application. This case will share a set of large-scale stateful cluster optimization methods through the interpretation of cruise control practice.
Low Cost System Interface Testing-Automation, Performance and Continuous Integration &; Online monitoring "
Jiu Hao ·DJI Test and Development Engineer
Case value: In most companies and projects, automatic testing, performance testing, continuous integration and online monitoring of system interfaces are needed. However, the existing methods all have some problems, such as low input-output ratio, too many tools and technology stack, and high maintenance and learning costs. In view of this common pain point, DJI explored a low-cost best practice scheme and precipitated it into the open source interface testing framework ApiTestEngine. This case will disassemble the technical points and implementation principle of the framework.
● Intelligent Operation and Maintenance @Pinterest
Meng Xiaoqiao-Manager of ——Pinterest Monitoring Department
Case value: Intelligent operation and maintenance is the future trend of all cloud-based companies. As a large-scale photo sharing platform, PINTEREST has a huge and complex background computing platform and software architecture. How to ensure high-quality operation and maintenance with the least human and resource costs is a huge challenge. Therefore, our monitoring department has built a comprehensive monitoring platform, which has three characteristics: high scalability, integration and intelligence. This case will provide practical exploration for intelligent operation and maintenance through sharing the monitoring platform.
The above are some selected cases. For more information and timetable of TOP 100 case, please visit [official website]. 4 days, focusing on sharing 20 17 100 R&D case practices worth learning. A total of 10 free tickets for the opening ceremony were issued on this platform. Limited quantity, first come, first served. Free experience coupon application entrance.