I think, therefore I am; recording insights from daily life and technology.
This article explores the critical role of design decisions in software architecture, introduces templates like ADR for documenting decisions, discusses challenges in extracting decision knowledge, and examines how LLMs can mine historical artifacts like GitHub issues to recover lost architectural rationale.
Notes from Jeff Dean's recent interviews reveal his engineering philosophy: build large models then distill to efficient flash models, design evaluation sets that expire at 95% saturation to guide architecture optimization, and implement long-context processing via parallel coarse-to-fine retrieval akin to search engines.
This article explores software engineering as a progression of abstraction levels — from functions and classes to libraries, platforms, and AI-driven design primitives — arguing that each leap releases productivity by encoding structural 'genes' that enable building more complex systems.
The article argues AI expands software capabilities for previously unsolvable problems but won't replace deterministic systems because accumulated business knowledge, reliability requirements, and ROI make rewriting unjustified; code agents generate deterministic code as a pragmatic compromise.
A veteran kernel engineer argues that while AI academia floods the field with "AI-for-kernel" papers, few reach mainline due to community conservatism and engineering gaps; instead, kernel developers should focus on "kernel-for-AI" optimizations—scheduling, memory, I/O for AI workloads—where real impact lies.
The article explores structural thinking as the core mindset of software engineers, breaking it into three levels—conception, insight, and understanding—while emphasizing runtime and evolutionary structures for managing complexity, enabling collaboration, and sustaining long-term system health.
The article examines software engineering's fundamental complexity laws, large language model limitations including context constraints and hallucination, Rice's Theorem implications, and questions the boundaries of AI agents in solving real software problems despite improved memory, tools, and context organization.
This technical briefing curates 14 deep-dive articles covering AI agent evaluation methodologies, real-world task benchmarking, Google engineering lessons, Rust vs C assembly performance, MCP tool calling internals, structured logging design, minimalist agent architectures, Codex agent internals, historical developer replacement attempts, observability evolution, vLLM memory leak debugging, hotspot optimization pitfalls, and LLM non-determinism root causes.
The article defines six core traits of engineering practice — efficiency, ambiguity, fragility, resilience, maintainability, and safety — each categorized into hard, soft, and chaotic types, illustrated with historical case studies and structured methods for managing them.
This article explores the engineering mindset through problem classification (hard, soft, messy), three core tools (structure, constraints, trade-offs), modular and reverse thinking, constraint-driven innovation, systems engineering across six dimensions, and multi-level visualization for complex problem solving.
