A comprehensive production engineering monograph structured after the classical Thirty-Six Stratagems across six volumes, 36 chapters, and 242 sections. Anchored to reproducible labs on PostgreSQL 18.6 and Pigsty 4.5, it walks from relational modeling, advanced SQL, MVCC concurrency, and index internals all the way to Patroni high availability, PITR backup, observability, cloud-exit economics, and incident postmortems.
Original Works & Publications
Original engineering monographs on production PostgreSQL, bilingual digital canons, peer-reviewed research papers, and interactive data reports.
A cross-civilizational bilingual digital canon bringing together foundational texts from seven traditions — Buddhism, Daoism, Confucianism, Christianity, Islam, Hinduism, and Western Philosophy — alongside architectural essays on LLM alignment, perception, memory, causality, and autonomous AI agent governance.
Research Papers
Peer-reviewed research papers on LLM-powered database root-cause diagnosis and heterogeneous data federation via PostgreSQL FDW.
Co-authored with Prof. Guoliang Li’s database group at Tsinghua University and ModelBest, published at VLDB 2024 (PVLDB Vol. 17, pp. 2514–2527). D-Bot extracts diagnostic chunks from operational manuals, equips domain-specialist LLM agents with metric/log/SQL inspection tools, and coordinates them via tree-search reasoning to pinpoint root causes of complex compound anomalies in production PostgreSQL clusters.
Published at the 14th IFIP International Conference on Network and Parallel Computing (NPC 2017, Springer LNCS Vol. 10578, pp. 173–185). Born out of production systems at Alibaba, this paper presents a Foreign Data Wrapper (FDW) federation layer that enables cross-engine joins, predicate pushdown, and ETL pipelines across HBase, Redis, MongoDB, and MySQL through standard PostgreSQL SQL.
Talks, Reports & Dossier
International conference slide decks, interactive developer survey visualizations, kernel patch reports, and PostgreSQL community nomination dossiers.
Interactive HTML slide deck presented at PGConf.Dev 2026 in Vancouver, Canada. Dissects the "last-mile delivery problem" of the PostgreSQL extension ecosystem and demonstrates how the pgext.cloud metadata registry, the PIG CLI package manager, and the SOW repository build pipeline make 570+ binary extensions installable with one command across 16 Linux distributions.
A single-page interactive data journalism report built from the StackOverflow Developer Survey dataset. Visualizes PostgreSQL’s multi-year surge to the #1 most used, admired, and desired database, alongside cloud provider trajectories, programming language shifts, and the adoption vs. trust gap in AI coding assistants.
A structured community nomination dossier documenting over a decade of independent work in the PostgreSQL ecosystem (2015–2026): maintaining 570+ binary extension packages and pgext.cloud, authoring the Pigsty open-source RDS distribution, translating the PostgreSQL 14–19 manuals and 10+ books, contributing kernel patches, and delivering 25+ conference talks.
An interactive visual analysis of the Anthropic Economic Index dataset, mapping millions of real-world Claude conversations onto the O*NET occupational taxonomy to measure automation vs. augmentation across software engineering, database operations, data analysis, and knowledge work.
An in-depth kernel bug analysis and patch report examining how PostgreSQL Range and Multirange types compute distances, build ANALYZE length histograms, and evaluate GiST index penalty functions when endpoints contain IEEE 754 NaN values — complete with reproducers and cross-version behavior matrices.
A high-density interactive single-page architecture board presenting the evolution of Pigsty since 2018, its four core modules (PGSQL, INFRA, NODE, ETCD), high-availability failover topology, Linux distro compatibility matrix, and full observability stack.