Build Your Own Git, Docker, and More from Scratch with Python, Rust, and C

I built a mini code judge from scratch — and it runs 1000 test cases in just a few seconds. 🚀 Here's how it started: I was curious — how do platforms like Codeforces and LeetCode actually work under the hood? So I decided to build one myself using nothing but Docker and a few shell scripts. Here's what I built: ✅ Pulled the official Rust Docker image ✅ Wrote a Rust program locally in Notepad ✅ Mounted my local folder into the container using volume mounts ✅ Compiled and executed the code entirely inside the container ✅ Built a judge script that feeds 1000 test inputs, captures outputs and compares them against expected results ✅ Tracked total script time end to end The result? 1000 test cases — compiled, executed, and verified in just 16 seconds. That includes compilation, 1000 program runs, 1000 file reads, and 1000 comparisons. What I learned: 🔹 Online judges are not magic — they are just sandboxed execution + input/output comparison 🔹 The same architecture extends to Python, Perl, C, Java and even SQL Started with a simple "Hello World" — ended up understanding how online judges work at a systems level. Sometimes the best way to understand a tool is to build it yourself. 💡 #Rust #Docker #Systems #Programming #Learning #OnlineJudge #DevJourney

3 𝐝𝐚𝐲𝐬. That's how long I spent debugging a deployment error that turned out to be a Python version mismatch Not a logic error. Not a broken API. A version number. I checked my code line by line. Rewrote entire functions. Googled the same error message 12 different ways. The fix? Two characters in a config file.😅 That moment taught me more than any lecture on software engineering: → Read the error message properly before you fix anything → Check your environment before you check your code → The simplest explanation is usually correct Real-world dev isn't glamorous. It's mostly debugging things that shouldn't be broken in the first place. But every painful bug makes you faster at the next one. What's the most embarrassing bug you've ever spent too long fixing?😭 #SoftwareDevelopment #DevLife #Nigeria #CSStudent #BuildInPublic

Build and Connect Your First MCP Server to OpenCode on Windows: How to build your own MCP server from scratch using Python, and how to make it available to an LLM like Qwen 3.5 in OpenCode. Continue reading on Generative AI » #genai #shared #ai

Checking Codeforces, then AtCoder, then LeetCode just to keep track of contests is a productivity killer. So I built CPAgent. It fetches upcoming contests from Codeforces, AtCoder, LeetCode, and CodeChef every day and syncs them straight to Google Calendar. Correct timezone. No duplicates. No manual effort. Built with Python 3.10, Pydantic v2, and the Google Calendar API. Repo is live and open source. 🔗 GitHub: https://lnkd.in/g7JxU8g3 #OpenSource #Python #CompetitiveProgramming #Automation #GitHubActions

Just got my first package merged into nixpkgs! 🎉 It's vcs2l (https://lnkd.in/eQQQdQSQ), a little CLI tool for managing multiple git repos at once (basically a maintained fork of vcstool). But the real thing I'm sitting with: the more I mess around with Nix, the more I see its DNA on basically every modern dev tool I love. Cargo with its lockfiles and "builds are pure functions" energy? Very Nix-y. uv bringing reproducible Python environments with actual lockfiles? That's the Nix philosophy in Python-land. And pixi (of which I'm a huge fan!) explicitly calling out Nix as inspiration for project-local, language-agnostic environments with lockfile guarantees? They're literally bringing Nix concepts to the scientific computing world! It's wild to realize that Nix has been quietly teaching us how to think about software supply chains for decades, and now the rest of the ecosystem is catching up. Makes the infamous "Nix learning curve" feel worth it. Anyway, stoked to be package 120,000-and-something in nixpkgs. If you're managing multiple repos, give vcs2l a spin! #NixOS #OpenSource #DevTools #ReproducibleBuilds #Python #Rust

I’ve been off Linkedin and Youtube for some time reducing distractions and focusing on new skills. Lately, I improved my Rust knowledge by building new projects and migrating old systems from Python and Go to Rust. Not an easy task but absolutely possible. Turned out you don’t need to be a memory control freak and can think on a higher level and focus on the task at hand. Speed is not the first priority for me but the correctness and error handling, as dealing with production critical systems doesn’t leave an option for outage and forces me to think about edge cases. Speed and a single binary for easy distribution is a nice bonus :)

📓 Just published a FREE production-ready FastAPI authentication system on Docker Hub! ✅ JWT tokens ✅ Role-based access control ✅ Argon2 password hashing ✅ Docker containerized ✅ Pull & run in 2 commands Perfect for devs who need auth FAST without building from scratch. 🔗 Docker Image: https://lnkd.in/eb7_dHKr 💻 Source Code: https://lnkd.in/eKshJ5gz What would you build with this? Comment below! 👇 #FastAPI #Python #Docker #Authentication #Backend #OpenSource #100DaysOfCode

Day 18: Why the "How" matters more than the "What" 🏗️ Today’s #30DaysOfCode challenge was a classic: The Palindrome Checker. While most would reach for a simple .reverse() method, I took a deeper dive into Linear Data Structures using JavaScript to implement a dual-stream logic: Stacks (LIFO): To handle the reverse flow of data. Queues (FIFO): To maintain the original sequence. The Engineering Takeaway: By comparing the "pop" of a stack against the "dequeue" of a queue, you aren't just checking a string—you’re managing data flow efficiency. As I scale my work in Data Engineering, these fundamentals are the difference between a script that just "works" and a system that is built to scale. Shoutout to my AI collaborator #Gemini for the technical sounding board as I bridge these concepts across my Java and JS projects. #DataEngineering #JavaScript #Algorithms #DataStructures #30DaysOfCode #TechCareer

Ever wondered why some developers swear by Jupyter notebooks? In my journey with Java and Node.js, I’ve realized something: exploratory programming changes how you think, not just how you code. Jupyter and IPython create a space where ideas can be tested instantly—no boilerplate, no waiting, just thinking in motion. Instead of writing full scripts upfront, you: • experiment in small chunks • visualize results immediately • iterate without friction That shift alone makes complex problems feel more approachable. Here’s what’s worked for me: 1. Use Jupyter for data exploration and quick visualizations 2. Lean on IPython for fast calculations and iterative testing 3. Combine code + notes to document your thought process as you go It’s less about tools—and more about developing a mindset of curiosity and rapid feedback. These tools didn’t just improve my workflow—they sharpened how I solve problems. Curious—how do you explore data in your projects? #DataScience #Programming

So I finally gave up and asked a stupid question to #Copilot on a Github repo with scarce documentation. The question was simple: what are the values a given option could take and where is the file in which this us defined? The repo is a very simple python lib and I could have found the answer in a couple minutes but I was a bit tired. The answer it provided with #Claude Haiku was correct but for the wrong reasons. It gave me an incorrect option name, didn't provide the values but the conversion factor between the possible values, and hallucinated the file and it's format. So this is the piece of shit of an assistant everybody brags about in code development? Damned, no wonder some people claim the bubble will burst and we're entering the age of tech debt. 😐 #ResistAIGen