Browse ZiCode posts, columns, and notes in reverse chronological order, covering programming, AI tools, local models, and engineering practice.
The seventeenth article in Thirty Years in IT and Me. In 2012, I left Baidu and joined Weibo as it was sprinting toward its IPO. Baidu's organizational atmosphere was starting to change, while Weibo stood at the intersection of mobile internet, social media, and big data platforms.
The sixteenth article in Thirty Years in IT and Me. In 2011, Baidu was still in the bright years of the Chinese internet. Big companies then were not only companies; they were large engineering schools where traffic, systems, search, ads, data, and internal technical discussion pushed programmers into bigger real-world engineering scenes.
The fifteenth article in Thirty Years in IT and Me. In 2010, I moved from a system-software company into Baidu. Baidu still had a strong engineering atmosphere then, and its internal Hi groups were full of technical discussion.
Before the 2026 World Cup round of 16 begins, here is a fan-style prediction for the semifinalists and champion, based on the bracket, team form, squad depth, and a little football instinct.
The first article in Programmer Craft in the AI Era revisits program = algorithms + data structures, and explains why requirements must first become state, rules, complexity, and acceptance examples before a developer can know what to ask AI to build.
The second article in Programmer Craft in the AI Era looks at queues, hash tables, trees, indexes, and logs as they appear in real systems, and explains why they determine system boundaries, performance, and failure modes.
The third article in Programmer Craft in the AI Era covers processes, threads, coroutines, event loops, and blocking calls, and explains why concurrency is not just wrapping code in async or adding more workers.
The fourth article in Programmer Craft in the AI Era discusses function calls, HTTP/RPC, message queues, event streams, and shared databases. The point is not choosing a fashionable technology, but clarifying coupling, transactions, failure, and observability boundaries.
The fifth article in Programmer Craft in the AI Era covers request and response schemas, error codes, idempotency, pagination, permissions, version compatibility, and contract tests. Before asking AI to write an API, write down the promise first.
The sixth article in Programmer Craft in the AI Era covers table design, indexes, transaction boundaries, locks, slow queries, and migrations. A database is not a universal JSON box; it is the hardest business-constraint layer in a system.
The seventh article in Programmer Craft in the AI Era covers caches, queues, and asynchronous jobs. A cache is not a magic accelerator, a queue is not a trash bin, and async jobs still need idempotency, retries, dead letters, batching, and backpressure.
The eighth article in Programmer Craft in the AI Era covers connection pools, batching, rate limiting, degradation, hot spots, cascading failures, and capacity estimation. Concurrency is not a single coding trick; it is pressure design across the whole call chain.
The ninth article in Programmer Craft in the AI Era explains how unit tests, integration tests, fixtures, contract tests, regression tests, and failure cases can constrain AI-generated code so the result is reproducible, reviewable, and maintainable.
The tenth article in Programmer Craft in the AI Era explains how logs, metrics, traces, audit records, and product events form an observability boundary, so AI-generated services do not become black boxes.
The eleventh article in Programmer Craft in the AI Era explains how to review, isolate, test, draw boundaries, delete code, and refactor in small steps so an AI prototype can become maintainable software.
The twelfth and closing article in Programmer Craft in the AI Era returns to computer fundamentals, systems thinking, product judgment, communication, and collaboration, and discusses which abilities AI will amplify.
The useful part of vibe coding is not quickly throwing together a pretty page. It is using real data to find a product shape fast, then leaving behind contracts, fixtures, tests, and handoff notes that engineering can actually pick up.
The fourteenth article in Thirty Years in IT and Me. Around 2009, GitHub began turning open-source collaboration from mailing lists, patches, and project homepages into a more visible and social daily workflow. In the same year, Weibo appeared, and information streams started reshaping the Chinese internet.