• Finally Recovered All Investment Losses - Mostly Due to Luck

    After several years of ups and downs, I have finally recovered all my previous investment losses on paper. Looking back, I have to be honest about what really happened - it was mostly luck, at least in my case.

  • AI Coding Assistants: Potential and Inconsistency

    Over the past few weeks, I’ve been experimenting with various AI coding assistants including Qwen, Gemini, Cursor, and Claude Code. My experience has been a mix of successes and setbacks.

  • Merging Frontend and Backend in a Single Cloudflare Worker

    On August 25, 2025, a significant step was made in the development of the surname project, focusing on unifying the frontend (OpenNext.js) and backend (Hono.dev) into a single Cloudflare Worker. This approach streamlines deployment and maintenance, moving away from separate deployments for each service.

  • AI Quota Disparity: GPT-5 vs DeepSeek

    Here is my recent experience with OpenAI’s GPT-5.

  • AI Agents and Custom Model Providers

    The past two days have been a deep dive into AI agents and custom model providers. My goal was to find an agent that offers flexibility in choosing LLMs, particularly for cost-effective options like DeepSeek, and leveraging the free tier of Gemini where possible. Here’s a rundown:

  • Experience with Cursor AI Assistant

    Today I used Cursor membership for several coding tasks, and the experience was good. Not only did it help me complete jobs efficiently, but it even identified a missing component in my previous work that I’d overlooked.

  • The Rise of OS-Level MCP Services

    From Microsoft’s Copilot to Apple’s Apple Intelligence, both Windows and macOS are integrating AI features at an unprecedented pace. Even hardware manufacturers like AMD are branding their products as “AI CPUs.” AI and Large Language Models (LLMs) are being integrated into many products.

  • Understanding MCP and the Agent-Led LLM World

    I’ve been hearing the term “MCP” (Model Context Protocol) floating around a lot lately, especially after my company’s AI Week. I saw presentations and even a colleague demonstrating how simple it was to spin one up. I heard that MCP is easy to use, but I didn’t have a good grasp of what it was.

  • Serverless: From Theory to Practice

    Several years ago, when AWS first introduced its serverless offerings, an article described how serverless functions enabled cloud providers like AWS to achieve unprecedented hardware utilization efficiency while allowing developers to completely offload scalability concerns to the platform. The idea of not having to predict traffic patterns or manually configure scaling rules was appealing.

  • Vibe Coding with AI

    Recently, our company invested in some AI services. While they’ve been generally good, I agree with DHH’s perspective: AI is a good learning tool, but when you let it handle tasks directly, you often end up cleaning up after it.

  • My First Website

    Of course I have websites built with wordpress for my friend’s law firm, or by other tools. This is the first website that I build from scratch by coding. I have worked on Node.js for BFF and React.js for Mobile but never for websites.

  • Cloudflare worker for fullstack

    Why

    Cloudflare workers are a powerful way to run services and websites, especially for SSG + CSR, which is a great solution for both SEO and performance. It also takes the burden off of devops and scale management, as Cloudflare handles the scaling and infrastructure. And it’s free for starter projects.