<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0">
    <channel>
        <title>Silver Bullet</title>
        <link>https://findns94.github.io</link>
        <description>Martin Zhao's Blog</description>
        <lastBuildDate>Mon, 27 Jul 2026 14:45:09 GMT</lastBuildDate>
        <docs>https://validator.w3.org/feed/docs/rss2.html</docs>
        <generator>https://github.com/jpmonette/feed</generator>
        <language>zh-CN</language>
        <copyright>Martin Zhao 2026</copyright>
        <item>
            <title><![CDATA[How to Find a Profitable Side Hustle as a Backend Developer in the AI Era: A 2026 Guide]]></title>
            <link>https://findns94.github.io/posts/backend-developer-side-hustle-ai-era/</link>
            <guid>https://findns94.github.io/posts/backend-developer-side-hustle-ai-era/</guid>
            <pubDate>Mon, 27 Jul 2026 23:00:00 GMT</pubDate>
            <description><![CDATA[![A developer working on a laptop in a modern workspace, symbolizing the solo builder economy in the AI era](/posts/backend-developer-side-hustle-ai-era/images/cover-laptop.jpg)

# How to Find a Profitable Side Hustle as a Backend Developer in the AI Era: A 2026 Guide

In 2024, developers using AI coding tools completed tasks **55% faster** on average, and **46% of all new code on GitHub was AI-generated** — up from roughly 25% the year before ([GitHub Octoverse 2024](https://github.blog/news-insights/octoverse/octoverse-2024/)). The implication is profound: a single backend developer can now build in weeks what previously required a team of three or four. Yet most backend engineers still trade time for money on the side — freelancing by the hour, picking up overtime — without recalibrating for a world where the bottleneck is no longer writing code, but finding the right problem to solve.

This guide provides a structured framework for backend developers who want to find, validate, and launch a profitable one-person project in the AI era. It is not about chasing hype. It is about applying your existing systems-thinking skills to the highest-leverage opportunities AI has created.

> **Key Takeaways**
> - AI coding tools have compressed development time by 55% on average and up to 12x for specific security fixes, making one-person products viable at a fraction of the historical cost ([GitHub Octoverse 2024](https://github.blog/news-insights/octoverse/octoverse-2024/)).
> - The global SaaS market is projected to reach $919.98 billion by 2030 (CAGR 17.7%), with AI-integrated products capturing a disproportionate share of new growth ([Grand View Research](https://www.grandviewresearch.com/industry-analysis/software-as-a-service-saas-market-report), 2024).
> - Only 2-5% of developer side projects reach $10K+ MRR, but the median time to first revenue for successful projects is just 3-6 months ([Indie Hackers](https://www.indiehackers.com), 2024).
> - Backend developers have a structural advantage: API-first thinking, data modeling skills, and infrastructure experience map directly onto AI-native product development.
> - The biggest risk is not failure — it is building something nobody wants. Pre-selling before coding is the single most effective de-risking tactic.]]></description>
            <category>AI</category>
            <category>Career</category>
            <category>Software Engineering</category>
        </item>
        <item>
            <title><![CDATA[How to Contribute to the Linux Kernel: A 2026 Guide for Amateur Engineers in the AI Era]]></title>
            <link>https://findns94.github.io/posts/kernel-contribution-guide-ai-era/</link>
            <guid>https://findns94.github.io/posts/kernel-contribution-guide-ai-era/</guid>
            <pubDate>Thu, 23 Jul 2026 23:30:00 GMT</pubDate>
            <description><![CDATA[![How to Contribute to the Linux Kernel - A 2026 Guide for Amateur Engineers in the AI Era, with kernel development workflow and AI-assisted coding statistics](/posts/kernel-contribution-guide-ai-era/images/cover.svg)

# How to Contribute to the Linux Kernel: A 2026 Guide for Amateur Engineers in the AI Era

In 2026, the Linux kernel 7.1 release pulled in 15,849 non-merge changesets from 2,479 developers — and 530 of them were first-time contributors ([LWN.net, "Who Wrote 7.1"](https://lwn.net/Articles/1077425), 2026). The kernel isn't a closed fortress. But its contribution process is famously unforgiving: plain-text email, strict style rules, and a review culture that assumes you've done your homework. If you've ever wanted to contribute but felt lost in the workflow, this guide is your map. You'll go from a fresh clone to a submitted, reviewable patch — and you'll learn how AI tools can accelerate every step without breaking the community's norms.]]></description>
            <category>Linux</category>
            <category>Kernel</category>
            <category>Open Source</category>
        </item>
        <item>
            <title><![CDATA[Trapped Behind the Firewall: The Future of IT Bottlenecks in Chinese State-Owned Enterprises and Institutions]]></title>
            <link>https://findns94.github.io/posts/chinese-soe-it-bottlenecks-llm-restrictions/</link>
            <guid>https://findns94.github.io/posts/chinese-soe-it-bottlenecks-llm-restrictions/</guid>
            <pubDate>Thu, 23 Jul 2026 23:00:00 GMT</pubDate>
            <description><![CDATA[![Trapped Behind the Firewall - The Future of IT Bottlenecks in Chinese State-Owned Enterprises and Institutions, with domestic AI adoption and digital transformation statistics](/posts/chinese-soe-it-bottlenecks-llm-restrictions/images/cover.svg)

# Trapped Behind the Firewall: The Future of IT Bottlenecks in Chinese State-Owned Enterprises and Institutions

In 2026, China's State-owned Assets Supervision and Administration Commission (SASAC) reported that over 98% of central state-owned enterprises have launched digital transformation initiatives, yet fewer than 15% have deployed large language models in production workflows ([SASAC](http://www.sasac.gov.cn/), 2026). The gap is not a technology problem — it is a policy constraint. Confidentiality requirements, data localization mandates, and national security regulations prevent SOEs and government institutions from sending sensitive data to external LLM APIs like OpenAI's GPT-5, Anthropic's Claude, or Google's Gemini. While the private sector races ahead with AI agents and autonomous workflows, China's most economically significant organizations face a fundamentally different bottleneck landscape.

This article examines how the inability to use advanced external LLM APIs is reshaping IT architecture, procurement priorities, and competitive dynamics for Chinese SOEs and public institutions — and what the future holds as domestic alternatives mature.

> **Key Takeaways**
> - In 2026, over 98% of China's central SOEs have launched digital transformation programs, but fewer than 15% have deployed LLMs in production due to confidentiality constraints ([SASAC](http://www.sasac.gov.cn/), 2026).
> - China's generative AI market is projected to exceed $1.5 trillion by 2030, with domestic models (DeepSeek, Qwen, ERNIE) capturing over 80% of the SOE and government segment ([IDC](https://www.idc.com/), 2025).
> - The "信创" (IT Application Innovation) policy mandates full domestic替换 of core IT infrastructure in government and SOE systems by 2027, creating a $50 billion replacement market ([CCID Consulting](https://www.ccidconsulting.com/), 2025).
> - SOE IT bottlenecks are shifting from hardware procurement to three new constraints: domestic GPU compute scarcity, proprietary data silos, and the talent gap in self-managed AI deployment.
> - Organizations that invest in private AI infrastructure and internal MLOps capabilities today will define the next decade of China's state-sector productivity.]]></description>
            <category>AI</category>
            <category>Enterprise Technology</category>
            <category>China Tech</category>
        </item>
        <item>
            <title><![CDATA[How to Ease Father-Daughter Tension in a Chinese Family: A Practical Guide (2026)]]></title>
            <link>https://findns94.github.io/posts/ease-father-daughter-relationship-chinese-family/</link>
            <guid>https://findns94.github.io/posts/ease-father-daughter-relationship-chinese-family/</guid>
            <pubDate>Thu, 23 Jul 2026 22:45:00 GMT</pubDate>
            <description><![CDATA[![Father and daughter enjoying a warm sunny day outdoors, representing hope for healing the Chinese father-daughter bond](/posts/ease-father-daughter-relationship-chinese-family/images/cover.jpg)

# How to Ease Father-Daughter Tension in a Chinese Family: A Practical Guide (2026)

In 2024, 74.6% of Chinese young adults reported communicating with their parents at least once a week, yet only 12.3% had a deep emotional conversation more than once a month ([China Youth Daily](https://www.chinayouth.cn/), 2024). That gap — between talking and truly connecting — is where most father-daughter tension lives. If your father carries himself like a old-school cadre (老干部): stiff, principled, allergic to vulnerability — and you've grown up with opinions he didn't ask for, you already know this feeling. The conversations that should be the warmest often feel like cross-examinations.

This guide won't tell you to "just talk to him." It offers something better: culture-aware, psychology-backed strategies that respect who he is while protecting who you're becoming.

> **Key Takeaways**
> - Chinese fathers and daughters talk often (74.6% weekly) but rarely go deep (12.3% meaningful talks monthly) — the gap is cultural, not personal ([China Youth Daily](https://www.chinayouth.cn/), 2024).
> - 81.3% of Chinese parents want emotional closeness with their children but say they "don't know how to start" ([Beijing Normal University Faculty of Psychology](https://psych.bnu.edu.cn/), 2024).
> - Side-by-side activities (walking, cooking, driving) outperform face-to-face talks for emotionally reserved fathers — shared action lowers the pressure to perform intimacy.
> - Healing doesn't require him to change first; small shifts in your approach can reshape the entire dynamic.]]></description>
            <category>Family</category>
            <category>Relationships</category>
            <category>Culture</category>
        </item>
        <item>
            <title><![CDATA[How to Create a Great Travel Vlog: A Beginner's Guide to Aesthetic Travel Video]]></title>
            <link>https://findns94.github.io/posts/how-to-create-great-travel-vlog-beginners-aesthetics/</link>
            <guid>https://findns94.github.io/posts/how-to-create-great-travel-vlog-beginners-aesthetics/</guid>
            <pubDate>Thu, 23 Jul 2026 14:30:00 GMT</pubDate>
            <description><![CDATA[![How to Create a Great Travel Vlog — a beginner's guide to aesthetic travel video, with key stats about travel video consumption](/posts/how-to-create-great-travel-vlog-beginners-aesthetics/images/cover.svg)

# How to Create a Great Travel Vlog: A Beginner's Guide to Aesthetic Travel Video

In 2026, 70% of travelers watch travel video content before they book a trip, and more than 280 million people engage with travel videos every month ([ThinkGoogle Travel Insights](https://business.google.com/en-all/think/), 2026). Yet most beginner vlogs look the same: shaky handheld clips, random transitions, no story. The gap between a forgettable clip and a beautiful travel vlog is not better gear. It is learning a handful of visual principles and applying them with intention. This guide walks you through the entire process — from planning to publishing — with a focus on the aesthetics that make travel video worth watching.

You do not need a cinema camera or a film degree. You need a clear story arc, a basic understanding of light and composition, and a simple editing workflow. By the end of this guide, you will have a repeatable system for creating travel vlogs that look polished and feel personal.

> **Key Takeaways**
> - In 2026, 70% of travelers watch travel video before booking, and more than 280 million people engage with travel videos monthly — the audience is there, but only well-crafted content earns the view ([ThinkGoogle Travel Insights](https://business.google.com/en-all/think/), 2026).
> - Aesthetic travel vlogs rest on four pillars: composition, color grading, intentional movement, and pacing. Master these before you buy expensive gear.
> - Audio matters more than video quality. Viewers forgive mediocre visuals; they click away from bad sound. Budget at least 25% of your gear spend on audio.
> - A three-act story arc (arrival, exploration, reflection) turns random clips into a narrative that holds attention past the first 30 seconds.]]></description>
            <category>Travel</category>
            <category>Video</category>
            <category>Photography</category>
        </item>
        <item>
            <title><![CDATA[Shanghai Property Won't Repeat Its 20-Year Boom — Here's What Japan, the US and the UK Actually Tell Us]]></title>
            <link>https://findns94.github.io/posts/shanghai-housing-market-long-term-trends-international-cycles/</link>
            <guid>https://findns94.github.io/posts/shanghai-housing-market-long-term-trends-international-cycles/</guid>
            <pubDate>Wed, 22 Jul 2026 22:30:00 GMT</pubDate>
            <description><![CDATA[![A wide aerial view of Shanghai's Pudong skyline and residential towers at dusk, representing the city's long-term housing market outlook](/posts/shanghai-housing-market-long-term-trends-internation]]></description>
            <category>Housing</category>
            <category>Finance</category>
            <category>Shanghai</category>
        </item>
        <item>
            <title><![CDATA[What Is an API Gateway? A Liberal Arts Student's Perspective]]></title>
            <link>https://findns94.github.io/posts/api-gateway-humanities/</link>
            <guid>https://findns94.github.io/posts/api-gateway-humanities/</guid>
            <pubDate>Sun, 19 Jul 2026 22:00:00 GMT</pubDate>
            <description><![CDATA[![What Is an API Gateway — A Liberal Arts Student's Perspective, showing an API gateway as a digital gatekeeper with adoption and security statistics for 2025](/posts/api-gateway-humanities/images/cover.svg)

# What Is an API Gateway? A Liberal Arts Student's Perspective

In 1947, the psychologist Kurt Lewin noticed something ordinary: the food that reaches a family's dinner table doesn't get there by accident. Someone — usually a housewife, in Lewin's framing — decides what passes through the kitchen "gate" and what doesn't. Lewin called this act "gatekeeping," and the idea reshaped how we understand power, media, and access. It traveled from dinner tables to newsrooms, where a 1950 study of a newspaper editor nicknamed "Mr. Gates" showed how one person's choices shaped what thousands of readers believed was important.

You don't need to write code to understand what an API gateway is. In fact, the conceptual tools you already carry from the humanities — ideas about translation, borders, power, and language — are precisely the right ones. In 2025, Postman's State of the API report found that 82% of organizations had adopted some form of API-first approach, with 25% going fully API-first ([Postman](https://www.postman.com/state-of-api/2025/), 2025). Almost every digital interaction you make now passes through a gatekeeper you've never seen. By the end of this essay, you'll know what an API gateway actually does — and you'll have a sharper lens on gatekeeping itself.

> **Key Takeaways**
> - An API gateway is the internet's gatekeeper: one entry point that controls, translates, and protects traffic (Lewin 1947).
> - In 2025, 98% of organizations faced API security problems (Salt Security, 2024).
> - Four jobs — authentication, rate limiting, translation, logging — map onto ideas you know: borders, bouncers, switchboards.
> - A gateway is never neutral; it's policy wearing the mask of infrastructure.]]></description>
            <category>API</category>
            <category>Software Architecture</category>
            <category>Technology</category>
        </item>
        <item>
            <title><![CDATA[Is The Witcher 3 the Greatest RPG of the 21st Century?]]></title>
            <link>https://findns94.github.io/posts/witcher-3-greatest-rpg-21st-century/</link>
            <guid>https://findns94.github.io/posts/witcher-3-greatest-rpg-21st-century/</guid>
            <pubDate>Thu, 16 Jul 2026 22:20:06 GMT</pubDate>
            <description><![CDATA[![The Witcher 3 — 50+ million copies sold, 260+ Game of the Year awards, Metacritic 93, the greatest RPG of the 21st century](/posts/witcher-3-greatest-rpg-21st-century/images/cover.svg)

# Is The Witcher 3 the Greatest RPG of the 21st Century?

As of 2026, eleven years after its debut, The Witcher 3: Wild Hunt has sold over 50 million copies worldwide and won more than 260 Game of the Year awards ([CD Projekt investor reports via Wikipedia](https://en.wikipedia.org/wiki/The_Witcher_3:_Wild_Hunt), 2026). Few games stay commercially relevant for a decade; fewer still set the benchmark an entire genre is measured against. So does the numbers back up the "greatest RPG" claim, or is it nostalgia talking? This article weighs the data, the design, and the legacy — and offers a verdict.

"Greatest" is a loaded word. So this review scores The Witcher 3 across five concrete dimensions: commercial success, critical acclaim, quest and narrative design, world and replayability, and lasting cultural influence. We also compare it honestly against the strongest rivals of the past decade — Baldur's Gate 3, Elden Ring, Skyrim, and Cyberpunk 2077.

> **Key Takeaways**
> - In 2026, The Witcher 3 has sold more than 50 million copies — one of the best-selling RPGs ever — and holds more than 260 Game of the Year awards, making it one of the most decorated games in history ([CD Projekt investor reports via Wikipedia](https://en.wikipedia.org/wiki/The_Witcher_3:_Wild_Hunt), 2026).
> - It scores 93 on Metacritic (PC), 94 on current-gen consoles, and 95% on OpenCritic — elite marks — while its Blood and Wine expansion alone scores 92, higher than most full games ([Metacritic](https://www.metacritic.com/game/pc/the-witcher-3-wild-hunt/), 2026).
> - Its side quests are widely considered the genre's gold standard: choice-driven, morally grey, and written to the same quality as many games' main stories. Personal playthrough confirms it rewards slow, attentive play.]]></description>
            <category>RPG</category>
            <category>Gaming</category>
            <category>The Witcher 3</category>
        </item>
        <item>
            <title><![CDATA[No AI Without API: How AI Agents Transform Enterprise Workflow and Improve Production Efficiency]]></title>
            <link>https://findns94.github.io/posts/ai-agent-enterprise-workflow-transformation-2025/</link>
            <guid>https://findns94.github.io/posts/ai-agent-enterprise-workflow-transformation-2025/</guid>
            <pubDate>Thu, 16 Jul 2026 21:31:42 GMT</pubDate>
            <description><![CDATA[![No AI Without API - How AI Agents Transform Enterprise Workflow and Improve Production Efficiency, with enterprise AI agent adoption and ROI statistics for 2025](/posts/ai-agent-enterprise-workflow-transformation-2025/images/cover.svg)

# No AI Without API: How AI Agents Transform Enterprise Workflow and Improve Production Efficiency

In 2025, the McKinsey Global Institute found that generative AI could add up to \$4.4 trillion in annual value to the global economy, with 72% of enterprises now using AI in at least one function ([McKinsey Global Institute](https://www.mckinsey.com/mgi/our-research/generative-ais-impact-on-productivity-at-scale), 2025). Yet the missing link between AI hype and real business outcomes is simple: without robust API infrastructure, AI agents cannot access the data, tools, or systems they need to act autonomously. The agentic AI revolution is fundamentally an API story.

AI agents — software systems that perceive environments, make decisions, and take actions to achieve goals — are reshaping how enterprises handle customer service, IT operations, software development, and back-office workflows. This article examines the concrete mechanisms through which API-connected AI agents transform enterprise workflows and deliver measurable efficiency gains.

> **Key Takeaways**
> - In 2025, 81% of enterprise leaders expect AI agents to be moderately or extensively integrated into company strategy within 12–18 months, and 46% already use agents to fully automate workflows ([Microsoft Work Trend Index](https://www.microsoft.com/en-us/worklab/work-trend-index), 2025).
> - McKinsey estimates generative AI could add \$2.6–\$4.4 trillion annually across 63 analyzed use cases, but only 1 in 3 organizations have scaled AI beyond pilot stage ([McKinsey](https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/the-state-of-ai), 2025).
> - Klarna's AI assistant handled 2.3 million conversations in its first month — equivalent to 700 full-time agents — reducing resolution time from 11 minutes to under 2 minutes ([Reuters](https://www.reuters.com/technology/klarnas-ai-assistant-handled-two-thirds-customer-service-conversations-2025-02-25/), 2025).
> - API-first architecture is the non-negotiable foundation: without programmatic access to enterprise systems, agents cannot execute transactions, query data, or orchestrate multi-step workflows.]]></description>
            <category>AI</category>
            <category>Machine Learning</category>
            <category>Automation</category>
        </item>
        <item>
            <title><![CDATA[Is Shanghai School District Housing Still Worth It? A Ten-Year Decision Guide for Young Couples (2026-2036)]]></title>
            <link>https://findns94.github.io/posts/shanghai-school-district-housing-2026/</link>
            <guid>https://findns94.github.io/posts/shanghai-school-district-housing-2026/</guid>
            <pubDate>Thu, 16 Jul 2026 20:09:03 GMT</pubDate>
            <description><![CDATA[![Shanghai city skyline with school building silhouettes, representing the intersection of school district housing and education policy](/posts/shanghai-school-district-housing-2026/images/cover.svg)

# Is Shanghai School District Housing Still Worth It? A Ten-Year Decision Guide for Young Couples (2026-2036)

If you're a young couple in Shanghai debating whether to drain six wallets to buy a school district home, what you're really facing is a probability problem—not the "buy it and win it" certainty bet of the past. In 2024, Shanghai recorded roughly 112,000 newborns ([Xinhua](https://www.xinhuanet.com/), 2024), nearly halved from the 2016 peak of 218,000. Meanwhile, teacher rotation covered 80% of compulsory education schools by 2025 ([Shanghai Municipal Education Commission](https://edu.sh.gov.cn/), 2025), municipal key high school quotas are allocated to about 60% of non-selective junior highs ([Shanghai Municipal Education Examination Authority](https://www.shmeea.edu.cn/), 2025), and multi-school zoning continues to expand. Under the weight of these three overlapping policies, school district housing is shifting from a "certainty premium" to a "probability premium"—you're paying 100% of the premium but no longer buying 100% of the certainty.

This article uses data and trends to answer one question: over the next ten years, is Shanghai school district housing still worth the money for young couples?

> **Key Takeaways**
> - Shanghai school district housing premiums have narrowed from a peak of 40–50% to 15–35%, and the compression is still underway (Beike Research Institute / CRIC, 2024–2025).
> - Primary school enrollment pressure has eased noticeably since 2025—the 2024–25 school year saw roughly 750,000 enrolled students, down 14% from the 2021 peak of 871,000 (Ministry of Education, 2025).
> - Teacher rotation covered 80% of compulsory education schools in 2025, with an annual rotation rate of no less than 10–15% and backbone teachers accounting for ≥20% (Shanghai Municipal Education Commission, 2025).
> - School district housing yields a gross rental return of just 1.2–1.5%, well below the 1.8–2.2% on non-district newer homes (China Index Academy, 2026).
> - Young couples with both owner-occupier and school district needs should prioritize newer homes in mid-tier school districts, reducing the policy and liquidity risks of old, small, run-down units.]]></description>
            <category>Housing</category>
            <category>Education</category>
            <category>Shanghai</category>
        </item>
        <item>
            <title><![CDATA[How to Educate Young Children in the AI Era at Shanghai: A 2026 Step-by-Step Guide]]></title>
            <link>https://findns94.github.io/posts/young-children-ai-education-shanghai/</link>
            <guid>https://findns94.github.io/posts/young-children-ai-education-shanghai/</guid>
            <pubDate>Wed, 15 Jul 2026 22:47:02 GMT</pubDate>
            <description><![CDATA[![A young Asian girl placing her hand on a white humanoid robot, representing early-childhood AI interaction in Shanghai](/posts/young-children-ai-education-shanghai/images/cover.jpg)

# How to Educa]]></description>
            <category>AI</category>
            <category>Education</category>
            <category>Shanghai</category>
        </item>
        <item>
            <title><![CDATA[How to Choose an Investment Strategy in China for the Next 10 Years: A 2026 Guide for Young Couples in Shanghai]]></title>
            <link>https://findns94.github.io/posts/how-to-choose-investment-strategy-in-china-for-young-couples-shanghai/</link>
            <guid>https://findns94.github.io/posts/how-to-choose-investment-strategy-in-china-for-young-couples-shanghai/</guid>
            <pubDate>Tue, 14 Jul 2026 22:32:23 GMT</pubDate>
            <description><![CDATA[![Dazzling night view of Shanghai's Pudong skyline with the iconic Oriental Pearl Tower reflecting on the Huangpu River](/posts/how-to-choose-investment-strategy-in-china-for-young-couples-shanghai/i]]></description>
            <category>Finance</category>
            <category>Investment</category>
            <category>Shanghai</category>
        </item>
        <item>
            <title><![CDATA[How to Choose a Shanghai House-Buying Strategy for Young Couples (2026–2036)]]></title>
            <link>https://findns94.github.io/posts/shanghai-house-buying-strategy-young-couples/</link>
            <guid>https://findns94.github.io/posts/shanghai-house-buying-strategy-young-couples/</guid>
            <pubDate>Tue, 14 Jul 2026 20:09:10 GMT</pubDate>
            <description><![CDATA[![A wide aerial view of downtown Shanghai's Pudong skyline and residential towers at dusk](/posts/shanghai-house-buying-strategy-young-couples/images/shanghai-skyline.svg)

# How to Choose a Shanghai House-Buying Strategy for Young Couples (2026–2036)
If you are a young couple weighing your first home purchase in Shanghai, you are working through the hardest property math in the country. By early 2026, Shanghai's second- hand price index had slipped roughly 8-12% from its 2021-2022 peak even as transaction volumes recovered ([National Bureau of Statistics](https://www.stats.gov.cn), [Shanghai Real Estate Brokers Association](http://www.shfdcrea.com), 2025–2026). Prices are no longer on a one-way bet up, and mortgage underwriting is tighter in real terms than it looks. At the same time, the city's hukou-linked benefits (school districts, housing-fund access, public services) still anchor long-term demand in a way that smaller cities cannot match. The question is no longer "buy everything now" versus "do nothing." The question is which purchase geometry — which district, what policy window, what mortgage structure, what fallback plan — makes sense for the next ten years of your life, starting from where you stand today.

> **Key Takeaways**
> - Shanghai rental yields hover around 1.4–1.8%, well below what you can earn on a simple money-market fund — so renting is a stronger pure-financial play than many couples assume ([JLL China Residential Report](https://www.jll.com/en-china), 2025).
> - The middle ring (Putuo, Hongkou, Yangpu, part of Pudong) offers the best mix of price liquidity, metro connectivity and livability for young couples wary of exurban lock-in.
> - Shanghai's first-home mortgage rate was roughly 3.50–3.60% by early 2026 (5Y LPR minus basis points), historically low by Chinese standards and a genuine window for well-qualified buyers.
> - The winning strategy for most couples is a 5–10 year committed "buy and live" plan in the middle ring, backed by a joint-housing-fund loan and a cash buffer that survives a six-month job-loss scenario.]]></description>
            <category>Housing</category>
            <category>Finance</category>
            <category>Shanghai</category>
        </item>
        <item>
            <title><![CDATA[How to Submit Your First Linux Kernel Patch: A 2026 Step-by-Step Guide]]></title>
            <link>https://findns94.github.io/posts/submit-linux-kernel-patch/</link>
            <guid>https://findns94.github.io/posts/submit-linux-kernel-patch/</guid>
            <pubDate>Tue, 14 Jul 2026 09:00:00 GMT</pubDate>
            <description><![CDATA[![A computer monitor displaying lines of colorful source code on a dark background](/posts/submit-linux-kernel-patch/images/cover-code-on-monitor.jpg)

# How to Submit Your First Linux Kernel Patch: A 2026 Step-by-Step Guide

In 2026, the Linux kernel 7.1 release pulled in 15,849 non-merge changesets from 2,479 developers — and 530 of them were first-time contributors ([LWN.net, "Who Wrote 7.1"](https://lwn.net/Articles/1077425), 2026). The kernel isn't a closed fortress. But its contribution process is famously unforgiving: plain-text email, strict style rules, and a review culture that assumes you've read the manual. If you've ever wanted to contribute but felt lost in the workflow, this guide is your map. You'll go from a fresh clone to a submitted, reviewable patch — the same path those 530 newcomers walked this year.]]></description>
            <category>Linux</category>
            <category>Kernel</category>
            <category>Open Source</category>
        </item>
        <item>
            <title><![CDATA[Using ftrace to Selectively Trace Parent Functions]]></title>
            <link>https://findns94.github.io/posts/fstrace-filter-parent-function/</link>
            <guid>https://findns94.github.io/posts/fstrace-filter-parent-function/</guid>
            <pubDate>Sun, 07 May 2023 20:40:54 GMT</pubDate>
            <description><![CDATA[# Using ftrace to Selectively Trace Parent Functions

## Background

Taking the code in the memory cgroup subsystem as an example, the `__memory_events_show` function has two call sites, namely `memory_events_show` and `memory_events_local_show`.

The code is as follows:]]></description>
            <category>Linux</category>
            <category>Debugging</category>
            <category>Kernel</category>
        </item>
        <item>
            <title><![CDATA[Linux Kernel Learning — Part 1]]></title>
            <link>https://findns94.github.io/posts/learn-linux-step-1/</link>
            <guid>https://findns94.github.io/posts/learn-linux-step-1/</guid>
            <pubDate>Sun, 05 Dec 2021 20:22:30 GMT</pubDate>
            <description><![CDATA[# Overview

![linux_kernel_map](/posts/learn-linux-step-1/images/LKM.svg)]]></description>
            <category>Linux</category>
            <category>Kernel</category>
            <category>OS</category>
        </item>
        <item>
            <title><![CDATA[Booting a QEMU Linux System Using a Cross-Compiled aarch64 Kernel and BusyBox]]></title>
            <link>https://findns94.github.io/posts/qemu-aarch64-linux-in-wsl/</link>
            <guid>https://findns94.github.io/posts/qemu-aarch64-linux-in-wsl/</guid>
            <pubDate>Sun, 12 Sep 2021 22:15:59 GMT</pubDate>
            <description><![CDATA[## Compile BusyBox to Generate _install

### Install Build Dependencies bison/flex

```shell
sudo apt-get install bison -y
sudo apt-get install flex -y
```]]></description>
            <category>Linux</category>
            <category>Emulation</category>
            <category>Arm</category>
        </item>
        <item>
            <title><![CDATA[Learning Arm-v8 Assembly — Environment Setup]]></title>
            <link>https://findns94.github.io/posts/learn-arm-assembly-language/</link>
            <guid>https://findns94.github.io/posts/learn-arm-assembly-language/</guid>
            <pubDate>Sun, 16 May 2021 20:25:50 GMT</pubDate>
            <description><![CDATA[# Environment Setup

- Install the compilation and debugging components
    - Since I only had an x86 environment with WSL on hand, I needed to use emulation to compile, run, and debug ARM programs
    - Referencing this [gist article](https://gist.github.com/luk6xff/9f8d2520530a823944355e59343eadc1), it is possible to run armv8 programs on an x86 environment. The steps are as follows:
        - Install on WSL:
            - Cross-compilation environment: `sudo apt-get install libc6-dev-arm64-cross gcc-aarch64-linux-gnu`
            - QEMU emulation environment: `sudo apt-get install qemu qemu-system qemu-user`]]></description>
            <category>Arm</category>
            <category>Assembly</category>
            <category>Debugging</category>
        </item>
        <item>
            <title><![CDATA[The Page Fault Process After Address Access in Arm-Linux]]></title>
            <link>https://findns94.github.io/posts/what-happens-after-access-address/</link>
            <guid>https://findns94.github.io/posts/what-happens-after-access-address/</guid>
            <pubDate>Sun, 21 Feb 2021 19:34:02 GMT</pubDate>
            <description><![CDATA[# Address Access Example

Taking the access to the user-space **virtual address** 0x0000007000003000 as an example, consider the following code.

```C
#include <sys/mman.h>
#include <stdio.h>

int main()
{
    unsigned long long *x = (unsigned long long)0x0000007000003000;
    int *p = (int*)mmap(0x0000007000003000, sizeof(unsigned long long) * 10, PROT_READ | PROT_WRITE, MAP_SHARED | MAP_ANONYMOUS, -1, 0);
    printf("before write, x = %llu\n", *x);
    *x = 1;
    printf("after write,  x = %llu\n", *x);
}
```

If the user-space virtual address `0x0000007000003000` is not mapped in advance via mmap with read and write permissions, a `Segmentation fault (core dumped)` will occur, because `0x0000007000003000` is not managed as a vma (virtual memory area) allocated through mmap or brk in the process's virtual address space, and thus the virtual address is not accessible at this point. The output after compilation is as follows.

```
before write, x = 0
after write,  x = 1
```

As can be seen, a new value was successfully written to the user-space virtual address `0x0000007000003000`.]]></description>
            <category>OS</category>
            <category>Memory</category>
            <category>Linux</category>
        </item>
        <item>
            <title><![CDATA[Data Mining and Knowledge Discovery on KKBOX Music Data]]></title>
            <link>https://findns94.github.io/posts/kkbox/</link>
            <guid>https://findns94.github.io/posts/kkbox/</guid>
            <pubDate>Fri, 21 Jun 2019 21:23:16 GMT</pubDate>
            <description><![CDATA[KKBOX is a leading music streaming service in Asia, boasting the world's most comprehensive collection of Asian pop music. This paper conducts a data mining analysis on the music dataset that KKBOX has provided to the Kaggle community. Through initial cleaning and analysis of the dataset, we propose four data mining–related questions: correlation analysis among features in the dataset, song clustering, user clustering, and predicting whether a user will listen to a particular song repeatedly within a month. We then select appropriate algorithms—such as the K-prototypes clustering algorithm, the t-SNE high-dimensional data visualization algorithm, and the LightGBM algorithm—to address these data mining problems, and finally present corresponding conclusions. In doing so, we explore the application of data mining algorithms to a real-world dataset, transforming knowledge from the data mining classroom into truly practical technology.]]></description>
            <category>Data Mining</category>
            <category>Recommendation</category>
        </item>
        <item>
            <title><![CDATA[Formal Verification of Smart Contracts]]></title>
            <link>https://findns94.github.io/posts/dao-validation/</link>
            <guid>https://findns94.github.io/posts/dao-validation/</guid>
            <pubDate>Mon, 22 Apr 2019 22:41:57 GMT</pubDate>
            <description><![CDATA[A smart contract is a code contract and algorithmic contract that will become a foundational technology of the future digital society. It utilizes protocols and user interfaces to complete all steps of the contractual process. This article summarizes the main technical characteristics of smart contracts as well as existing trustworthiness and security issues, and proposes applying formal methods to the modeling, model checking, and formal verification of smart contracts to support the generation of large-scale smart contracts.]]></description>
            <category>Blockchain</category>
            <category>Security</category>
            <category>Formal Verification</category>
        </item>
        <item>
            <title><![CDATA[Transfer Learning for Face Recognition]]></title>
            <link>https://findns94.github.io/posts/face/</link>
            <guid>https://findns94.github.io/posts/face/</guid>
            <pubDate>Wed, 03 Apr 2019 22:49:31 GMT</pubDate>
            <description><![CDATA[The research history of face recognition is quite long-standing. As early as 1888 and 1910, Galton published two articles in *Nature* on using faces for personal identification, analyzing the human ability of face recognition. However, at that time, the problem of automatic face recognition was beyond reach. In recent years, face recognition research has attracted the attention of many researchers, and a variety of technical methods have emerged. In particular, since 1990, face recognition has made significant progress. Almost all well-known universities of science and engineering and major IT companies have research groups working on related studies.

In the early stages, traditional face recognition was usually studied as a general pattern recognition problem. The main technical approaches adopted were geometric feature-based methods. This was largely reflected in the study of profile silhouettes, where a great deal of research was devoted to the extraction and analysis of structural features from facial silhouette curves. Subsequently, appearance-based modeling methods such as Eigenface, Fisherface, and elastic graph matching were continuously proposed. Starting from the late 1990s, researchers began to focus on face recognition under real-world conditions, proposing different face space models, including linear modeling methods represented by Linear Discriminant Analysis, nonlinear modeling methods represented by kernel-based methods, and 3D face recognition methods based on 3D information. New feature representations were proposed, including local descriptors (Gabor Face, LBP Face, etc.) and deep learning methods.

Since 2014, deep learning + big data (massive labeled face data) has become the mainstream technical approach in the field of face recognition. Deep neural networks such as VGGFace, DeepFace, and FaceNet have been continuously proposed, and face recognition accuracy has been steadily improving. In 2014, Facebook's work DeepFace, published at CVPR 2014, combined big data (4 million face images) with deep convolutional networks, approaching human-level recognition accuracy on the LFW dataset. Google's work FaceNet, published at CVPR 2015, surpassed human-level recognition accuracy on the LFW dataset by adopting the Triplet Loss function.]]></description>
            <category>Deep Learning</category>
            <category>Computer Vision</category>
        </item>
        <item>
            <title><![CDATA[P2P Lending Industry Risk Analysis Platform]]></title>
            <link>https://findns94.github.io/posts/risk/</link>
            <guid>https://findns94.github.io/posts/risk/</guid>
            <pubDate>Thu, 28 Mar 2019 21:36:21 GMT</pubDate>
            <description><![CDATA[The goal of this project is to perform risk analysis for the P2P Internet finance industry, establish standardized risk assessment dimensions and risk assessment models, identify high-risk P2P Internet finance enterprises, and ultimately generate automated risk analysis reports for P2P Internet finance enterprises.

This project selected over 800 representative enterprises in the P2P Internet finance industry. Based on massive historical data from multiple aspects and dimensions of these enterprises, appropriate risk assessment dimensions were chosen and risk assessment models were built to analyze the risks currently faced by these P2P Internet finance enterprises. The overall risk profile and risk profiles across different domains of each enterprise were evaluated, and risk reports for P2P Internet finance enterprises were generated, providing an intuitive and accurate display of risk data, risk indicators, and risk ratings across all aspects of the enterprises. This meets the regulatory needs of Internet finance supervision and facilitates risk monitoring of Internet finance enterprises.]]></description>
            <category>Machine Learning</category>
            <category>Finance</category>
        </item>
        <item>
            <title><![CDATA[Blockchain Network Simulation]]></title>
            <link>https://findns94.github.io/posts/blockchain/</link>
            <guid>https://findns94.github.io/posts/blockchain/</guid>
            <pubDate>Tue, 19 Mar 2019 21:14:51 GMT</pubDate>
            <description><![CDATA[The simulation is divided into 3 parts:

- **P2P Network Simulation**
  - Use Mininet to build networks with different topologies (such as star, ring, tree, and mesh), and limit the bandwidth of different links.
  - Each node can generate certain types of data, communicate with other nodes, and build a local database that records the addresses of different nodes and the data types of each node.
  - Transmit data (of different sizes and in varying quantities) between different nodes.

---

- **Blockchain Simulation**
  - Implement a simulation of the blockchain network.
  - Simulate transactions (blocks) of the P2P network and the propagation logic.
  - Implement the PoW algorithm to simulate mining.
  - Limit link bandwidth and test the network latency and forking conditions under different block sizes and block generation intervals.
  - Use the PBFT protocol for consensus.

---

- **Attack Simulation**
  - Conduct experiments on the basis of the simulated Bitcoin network.
  - Run simulation experiments and record the probability of a successful attack (being able to produce a longer chain that invalidates certain existing blocks) under different hashing power levels.
  - BGP hijacking and Eclipse attacks.

The project repository is located at https://github.com/131250106/bitcoin]]></description>
            <category>Blockchain</category>
            <category>Security</category>
        </item>
        <item>
            <title><![CDATA[Hidden Markov Language Model]]></title>
            <link>https://findns94.github.io/posts/hmm/</link>
            <guid>https://findns94.github.io/posts/hmm/</guid>
            <pubDate>Tue, 19 Mar 2019 20:13:07 GMT</pubDate>
            <description><![CDATA[This post is a summary of the 5th assignment for the Fall 2018 Computational Linguistics course, with the task of using a `Hidden Markov Model` to calculate sentence probabilities.]]></description>
            <category>NLP</category>
            <category>Machine Learning</category>
        </item>
        <item>
            <title><![CDATA[Generating Adversarial Input Sequences Based on RNN]]></title>
            <link>https://findns94.github.io/posts/rnn-adversarial/</link>
            <guid>https://findns94.github.io/posts/rnn-adversarial/</guid>
            <pubDate>Sat, 09 Mar 2019 23:14:03 GMT</pubDate>
            <description><![CDATA[This article is a summary of the 6th assignment for the Fall 2018 Computational Linguistics course, with the task of generating adversarial input sequences based on RNN.

First, let me give an overall introduction to the concept of adversarial example attacks. Some researchers have found that although deep neural networks achieve very high accuracy, applying small perturbations to the input can cause the model's prediction results to be completely wrong. For images, such perturbations are usually so subtle that they are imperceptible (humans cannot intuitively perceive the perturbation in the image), yet these adversarial input examples can successfully fool deep learning models. This type of attack that deceives neural network models can be broadly divided into two categories: untargeted attacks only need to make the predicted result of the image inconsistent with the original label, while targeted attacks need to misclassify the image as a specific class. It is worth noting that researchers have found that generated adversarial examples are still effective on other models, which is known as the transferability of adversarial examples.]]></description>
            <category>Deep Learning</category>
            <category>NLP</category>
            <category>Security</category>
        </item>
        <item>
            <title><![CDATA[Multithreaded Concurrency Defect Detection for JVM Applications]]></title>
            <link>https://findns94.github.io/posts/confu/</link>
            <guid>https://findns94.github.io/posts/confu/</guid>
            <pubDate>Thu, 28 Feb 2019 23:27:41 GMT</pubDate>
            <description><![CDATA[Currently, with the rapid development of multi-core computer hardware, multithreaded concurrent programs are gaining increasing popularity and widespread adoption. However, due to the hidden nature of shared memory space access and the randomness of concurrent thread scheduling, multithreaded programs are highly susceptible to concurrency defects, which are often difficult to detect and reproduce. Existing concurrency defect detection methods all have their limitations; for example, methods based on system testing and symbolic execution cannot scale to large concurrent programs due to state space explosion, while methods based on probabilistic scheduling have a very low defect hit rate.

To address the enormous challenges faced in concurrent program testing today, the DATE team developed DATE-Confu, a multithreaded concurrency defect detection tool for JVM applications based on dynamic program testing techniques. The tool takes JVM executable programs under test (.class files or .jar files) as input and outputs the set of concurrency defects detected in the program. Currently, the tool can detect six categories of serious concurrency defects, including data races, deadlocks, and null pointer dereferences. Users can select which types of concurrency defects to detect based on their needs.

The key innovation of DATE-Confu lies in its effective combination of guided schedule fuzzing techniques and symbolic trace analysis techniques. The figure below shows the architecture diagram of DATE-Confu. The tool uses fuzzing techniques to efficiently traverse the enormous state space of multithreaded programs, while the symbolic trace analysis techniques rapidly discover unexplored control branches in the program under test, thereby providing effective information for iterative fuzzing and enabling DATE-Confu to reach higher coverage faster. DATE-Confu has so far been tested on multiple industrial-level projects. The experiments conducted to date have demonstrated that DATE-Confu achieves high testing efficiency and is capable of uncovering deeper-hidden concurrency defects.]]></description>
            <category>Concurrency</category>
            <category>JVM</category>
            <category>Testing</category>
        </item>
        <item>
            <title><![CDATA[Personalized Resume Recommendation]]></title>
            <link>https://findns94.github.io/posts/recommendation/</link>
            <guid>https://findns94.github.io/posts/recommendation/</guid>
            <pubDate>Sat, 23 Feb 2019 20:59:16 GMT</pubDate>
            <description><![CDATA[This system employs machine learning methods and a big data platform to train on a massive volume of job seekers' resumes. Unlike ordinary conditional filtering, this system can efficiently and accurately provide suitable and reliable talent recommendations from a big data processing perspective for IT positions that recruiters are looking to fill, reducing recruitment costs and using big data technology to bridge the gap between recruiters and outstanding talent.

Through research on massive resume data and company recruitment information, this project analyzes the characteristic features of personal resumes, companies, and company positions. From the recruiter's perspective, it conducts exploratory research centered on personalized recommendation technology to help recruiters obtain talent information in a more accurate and efficient manner. Based on this concept, a prototype system is implemented that can automatically recommend more appropriate resumes based on the characteristics of job postings.]]></description>
            <category>Data Mining</category>
            <category>Recommendation</category>
        </item>
        <item>
            <title><![CDATA[Detecting Improper Sitting Posture with a Laterally Positioned Motion-sensing Camera]]></title>
            <link>https://findns94.github.io/posts/sitting_posture/</link>
            <guid>https://findns94.github.io/posts/sitting_posture/</guid>
            <pubDate>Fri, 22 Feb 2019 20:00:00 GMT</pubDate>
            <description><![CDATA[**Abstract** Sitting posture detection is helpful for preventing musculoskeletal disorders. With the development of motion-sensing cameras and related software development kits (SDKs), it is possible to implement an application using skeleton detection technology. In this paper, a method is introduced to detect sitting posture from a lateral view without disturbing the user. To analyze video stream information, a skeleton thinning algorithm is described, and an averaging process is used to specifically locate the main joints from the lateral side. The results show this method has high accuracy when detecting improper sitting postures.
**Keywords** Ergonomics; Motion-sensing Camera; Gesture Recognition; OpenNI]]></description>
            <category>Deep Learning</category>
            <category>Computer Vision</category>
            <category>Health</category>
        </item>
    </channel>
</rss>