How ICBC’s AI Investment Advisor Reached Leading DevOps Continuous Delivery

According to a McKinsey study, technology companies succeed through standardization and tool empowerment, a principle that aligns with DevOps standards and continuous‑delivery pipelines.

On December 26, 2019, the Operations Conference was held in Beijing, organized by the China Information and Communication Research Institute and the China Communications Standardization Association, with support from the Cloud Computing Standardization Committee.

The conference announced the sixth batch of DevOps standard continuous‑delivery assessment results. China Industrial and Commercial Bank (ICBC) Smart Investment (AI‑投) project passed the third‑level assessment, earning a DevOps continuous‑delivery maturity certificate, a level considered leading in China.

Interview Highlights

Q: Please introduce yourself and the project. A: ICBC’s Software Development Center, founded in 1996, operates R&D teams in seven cities and handles application development, technology research, system training, and production operations.

The Smart Investment project leverages big data and artificial intelligence to provide personalized asset‑allocation services via mobile banking and smart terminals, serving millions of users.

Q: How did the project decide to join the DevOps assessment? A: As a flagship product of the bank’s smart‑banking strategy, rapid market response was essential, prompting the team to adopt DevOps practices and build a dedicated toolchain.

Q: What does passing the assessment mean for the team? A: It validates the team’s continuous‑delivery capabilities and confirms that their DevOps toolchain meets industry‑leading standards, boosting overall development efficiency.

Q: How does the system’s architecture support continuous delivery? A: The system is built on a fully cloud‑native architecture using ICBC’s proprietary IaaS and PaaS platforms. All nodes are deployed in the cloud, providing high availability, elastic scaling, fine‑grained logging, automated intelligent cloud operations, secure isolation, and end‑to‑end DevOps support. A front‑back separation with distributed service, log, monitoring, batch, and file centers enables sub‑second response times (<0.2 s).

Q: Describe the feature‑branch model. A: Features are linked to branches, allowing flexible release and rollback, and enabling rapid delivery of business requirements.

Q: Explain the commit‑build pipeline. A: After code submission, SonarQube scanning and unit tests run automatically within five minutes, enforcing code quality and test‑driven development.

Q: Explain the continuous‑delivery pipeline. A: The pipeline integrates SonarQube scanning, unit, smoke, and functional automated tests, providing gate‑controlled releases and strict version quality assurance.

The global first DevOps standard, the "Research‑Operations Integration (DevOps) Capability Maturity Model," was jointly created by the China Information and Communication Research Institute, cloud‑computing alliances, industry experts, Google, BATJ, Tsinghua University, Nanjing University, and leading enterprises. It is now officially listed by ITU‑T and the China Communications Standardization Association, with many Chinese banks and tech companies already certified.