[MM’s] Engineering Weekly #18 — July 28th, 2025
Exploring the future of coding, Java’s staying power, and AI’s unpredictable edge.
This week: AI’s impact on coding, Java performance wins, & timeless engineering principles.
AI, Coding Agents & the Future of Work
- Comparing Claude Code vs OpenCode (and testing different models): “A practical comparison of Claude Code and OpenCode (using Sonnet-4, Gemini Pro 2.5, and GPT-4.1), evaluating their reliability, code generation quality, and integration with developer workflows based on a real-world task.”
- Essential Reading for Agentic Engineers: “A curated collection of must-read articles and videos for mastering Claude Code, agentic coding workflows, and the future of AI-assisted development.”
- How Anthropic teams use Claude Code: “Discover how Anthropic’s internal teams leverage Claude Code for development workflows, from debugging to code assistance.”
- Improving the prompt to the AI to get better code: “After one-shoting the same prompt on multiple AI, I have created a refined prompt based on the various concerns with previous results.
- Nobody Knows How To Build With AI Yet: “The future of software development might just be jazz. Everyone improvising. Nobody following the sheet music.”
- No One Knows Anything About AI: “I want to present you with two narratives about AI. Both of them are about using this technology to automate computer programming, but they point …”
- The vibe coder’s career path is doomed: “Let me get one thing out of the way immediately: LLMs are helpful. This isn’t about whether LLMs can write code. They can. It’s about why vibe coding might be the worst career investment you’ll make.”
- Spending Too Much Money on a Coding Agent: “On making use of large thinking models.”
- What LLMs Know About Their Users: “How much the LLM — and the company — knows about its users. It’s a big quote, but I want you to read it all.”
- AI coding tool wipes production database, fabricates 4,000 users, and lies to cover its tracks: “An AI coding assistant from Replit has gone rogue, wiping a database and fabricating thousands of fake users.”
Java: Still Strong in 2025
- How I Made My Java Service 70x Faster by Rethinking JSON Deserialization: “We were running a production-grade Java microservice that handled thousands of requests per minute. It received API calls from multiple client apps, each carrying large, nested JSON payloads.”
- I Reviewed 500 Pull Requests — Here’s What Every Java Dev Gets Wrong: “From Optional disasters to stream overkill, these are the Java sins haunting your code reviews.”
- Speeding Up Spring Integration Tests: “Lessons from Context Caching.”
- Why Java is Still Worth Learning in 2025: “A Developer’s 25-Year Journey.”
Software Craft & Philosophy
- How to Think About Time in Programming: “A conceptual model for thinking about time in programming that encapsulates the complexity that many programmers cite online.”
- TODOs aren’t for doing: “Some teams require that every TODO comment in a codebase gets logged in the bug tracker. Others automatically delete any “stale” TODO that has been in the codebase for over a year.”
- SQL Injection as a Feature: “How We Opened Our Database to the World.”
- Work-Life Balance Slows Careers (E9 Engineer, ex-Meta): “Working harder DOES get you there faster — at a cost.”
Web & Tools
- Why your website should be under 14kB in size: “Having a smaller website makes it load faster — that’s not surprising.”
- How to Firefox: “Chrome finally pulled the trigger on the web’s best ad-blocker, uBlock Origin. Now that Chrome has hobbled uBO, Firefox — my beloved — 1 is surging again.”
Originally posted on marconak-matej.medium.com.