I think, therefore I am; recording insights from daily life and technology.
The author documents building a health check-in Android app using Vibe Coding with Trae and Cursor, comparing tool effectiveness, overcoming compilation issues, and reflecting on AI-assisted development's future impact on app creation.
The article argues that in a future where AI generates apps on demand, three core capabilities are essential to manage inherent uncertainty: automated testing from functional specs, runtime safety sandboxes for isolation and permissions, and AI-driven UI generation that replaces traditional UX design, enabling self-compiling, self-verifying, self-sandboxed, self-observing, and self-destructing software.
The article outlines three evolution directions for deterministic software: AI-friendly interaction redesign (UI/API for agents), runtime optimizations for inference engines, and adding adaptiveness via AI calls within secure sandboxes, noting architecture changes will be incremental via protocol-based adaptation layers.
Microsoft researchers propose self-evolving computing infrastructure that eliminates hard-coded redundancy, enabling systems to dynamically adapt and avoid loading 1-2GB of resources for a simple Hello World program.
This briefing curates ten pivotal articles on AI system engineering, covering agentic coding pitfalls, Anthropic's Skills framework, AI-friendly tech stacks, first-principles analysis of agentic coding, architecture diagramming, on-device Android AI deployment, code collaboration tool evolution, engineer mindset in the AI era, the Ralph Loop agent paradigm, and Cursor's multi-agent coding experience.
The article explores how software engineers can develop effective mental representations—covering architecture, concepts, temporal flow, spatial structure, decision knowledge, historical evolution, concerns, features, and patterns—through deliberate practice, abstract thinking, visualization, and inductive/deductive reasoning to enhance code understanding and expertise.
The article contrasts rule learners who extract abstract principles from examples with case learners who rely on analogies from past experiences, arguing that advanced structure builders who internalize underlying principles can better apply knowledge to unfamiliar situations, and reflects on the author's need to strengthen rule-based learning in software development.
An OSDI 2025 paper introduces a structured framework for serial performance optimization based on three principles—Removal, Replacement, Reordering—eight validated methodologies, and an AI model called SysGPT that provides optimization suggestions to developers.
The author reflects on code taste in the AI era, sharing three core principles from 'Code Complete 2' and 'Hackers & Painters': code form should mirror the problem, programmers should think before expressing in a language, and interfaces have both programmatic and semantic parts.
The article reflects on AI's transformative impact on software engineering, arguing that human value shifts from direct labor to enhancing AI capabilities. It outlines four AI-driven software paradigms (Agent/AI software, AI4SE, AI4System, SE4AI), predicts a hybrid future where agents orchestrate and traditional software handles determinism, and suggests three actionable paths for senior engineers: restructure legacy systems with AI, replace heuristic algorithms with LLMs, and fuse knowledge engineering with AI to create living digital experts.
