TODO-APP | Phase 1 | In-Memory Python Console Application | Hackathon 2 (GIAIC Quarter 4) Sharing my Hackathon 2 – Phase 1 project, where the objective was to design and implement an in-memory, console-based Todo application using a spec-driven and agentic development workflow. This phase focused on building a clean and predictable foundation before moving toward full-stack and AI-enabled phases. Key highlights: • Command-line Todo application with in-memory storage • Basic functionality implemented: add, view, update, delete, and mark tasks complete • Clean three-layer architecture (Models, Services, CLI) • Predictable in-memory behavior with validation for common user input errors • Spec-first, agentic workflow using Spec-Kit and Claude Code • Architecture aligned to support future full-stack and AI-based extensions GitHub Repository: https://lnkd.in/dxGDu6Em Looking forward to continuing this project in Phase 2. #GIAIC #GovernorSindhITInitiative #Hackathon2 #Phase1 #TodoApp #Python #CleanArchitecture #AgenticDevelopment

New Project: Building a Logic-Driven CLI Game in Go 🛠️ I’ve just published a new repository featuring a Number Guessing Game built entirely in Go. This project was a deep dive into Go’s fundamental concepts, focusing on dynamic randomisation, user input handling, and efficient state tracking. Key Technical Highlights: - Language: Go - Core Logic: Leveraged math/rand for target generation and fmt.Scanln for real-time user interaction in the terminal. - State Management: Used the slices package to track guess history, preventing duplicate entries without penalising the user. - Architecture: Built around a seamless infinite for loop that manages the game state until the win condition is met. One of the most rewarding parts of this project was navigating the transition from Python to Go. Specifically, learning how Go handles utility functions via modules rather than built-in operators (like Python’s in keyword) gave me a much deeper appreciation for Go’s approach to memory and types. Try it out yourself: 1️⃣ git clone the repo 2️⃣ go build -o guessing_game 3️⃣ ./guessing_game Check out the full source code and documentation here: 👉 https://lnkd.in/en3JzUyy #GoLang #OpenSource #SoftwareEngineering #GitHub #BackendDevelopment #Coding

I built a basic calculator using SpecKit Plus + Claude Code, and it feels fundamentally different from vibe coding. SpecKit Plus is a spec-driven development toolkit that puts a clear specification at the center of the workflow, guiding how AI agents plan, structure, and implement code rather than guessing based on loose prompts. It gives you templates and patterns so you can define what you want the system to do before code is generated. Here is how this approach helps: You start with a precise spec of behavior and requirements instead of ad-hoc prompts. The process produces predictable and consistent outcomes rather than relying on iterative prompting. It makes debugging and extending the code easier because the intent is documented up front. This is not vibe coding, where you keep asking for code until it works. This is spec-driven, structured, and repeatable. Below is a short demo of the calculator UI in action. Special thanks to Panaversity | Muhammad Junaid Shaukat | Wania Kazmi for the fascinating classes on #AgenticAI, #ClaudeCode, and #AINativeDevelopment. My GitHub Repo: https://lnkd.in/dVPVTs-4 #SpecDrivenDevelopment #AIcoding #Python #SpecKitPlus #SoftwareEngineering #Panaversity

The rule was straightforward: eat the dot, grow longer, and avoid colliding with yourself. Years later, while solving a Linked List problem on LeetCode, I had an unexpected realization. I had seen this before. The snake’s behavior was a perfect visual representation of a Linked List in action: A head moving forward A sequence of blocks following in order Dynamic growth as new elements are added Failure when the head intersects its own body What once felt like a childhood game was actually demonstrating a core data structure used in computer science. LeetCode didn’t introduce me to the concept from scratch. It simply gave terminology to something I had already experienced intuitively. It’s fascinating how early experiences, even simple games, shape the way we understand complex ideas later in life. From playing Snake for fun… to implementing and solving Linked List problems with code. A small reminder that learning often begins long before we recognize it. #LeetCode #DataStructures #LinkedList #Python #CodingJourney

I recently watched my colleague dive into a new Python project using GitHub Copilot - not just for quick code snippets, but as an active coding partner. What surprised me was how Copilot didn’t just offer boilerplate; it actually generated async functions with docstrings and thoughtful error handling. Sure, she still had to review and refine those suggestions (AI isn’t magic, after all), but more often than not, Copilot sped up the process - especially during those “blank editor” moments. 🚀 What stood out most: Copilot’s nudges towards better patterns, like security checks and best-practice usage of APIs, popping up right in the editor when she needed them. If you’ve seen GitHub Copilot surprise you (or a colleague), how do you make sure those suggestions are reliable? Any favorite tricks for reviewing what Copilot writes? Let’s swap stories! https://msft.it/6049tNvXv #GitHubCopilot #AIPairProgramming #CodeQuality #Python #JavaScript

Ever wished you could undo a coding mistake like Ctrl + Z for real life? 👀💻 Meet Git, the silent superhero behind every smart developer’s workflow. From tracking changes to fixing bugs without fear, Git turns mistakes into learning moments. This week’s #TechHackOfTheWeek is all about coding smarter, not safer. 🚀 Because in tech, it’s not about never breaking code, it’s about knowing how to fix it. #Technology #Python #HindustanCampus #ShardaGroup

Just Launched: Pydantic Helper & Snippets for VS Code I’m excited to share my new VS Code extension, Pydantic Helper & Snippets, designed to help Python and FastAPI developers write clean, correct Pydantic v2 code more efficiently. The extension provides practical snippets and command-based helpers for: Pydantic v2 models and validators Request & response schemas FastAPI endpoints and common patterns It’s lightweight, offline-friendly, and aligned with official best practices — no deprecated APIs, no hidden behavior. This project reflects my ongoing focus on building developer-friendly tools alongside my work at darsh fun, where I create browser-based games and practical web utilities. If you work with Python, FastAPI, or Pydantic, I’d love for you to try it and share feedback! Available on the VS Code Marketplace Support & feedback always welcome #Python #FastAPI #Pydantic #VSCode #DeveloperTools #OpenSource #Productivity

Unpopular opinion: Stop using LangChain for production apps. Don't get me wrong. It is an incredible tool for prototyping. If you need to build a demo in 30 minutes for a hackathon, use LangChain. But the moment you move to production, the "Abstraction" becomes a "Distraction." I spent 3 days last week trying to debug a Chain that was failing silently. I dug through 15 layers of wrapper code just to find out it was a simple API timeout. I rewrote the entire logic in raw Python using the OpenAI SDK. —> It took 50 lines of code. —> It was readable. —> It was debuggable. —> I knew exactly what was happening. We are developers. We shouldn't be afraid of if/else statements. We don't need a heavy framework to concatenate two strings and send them to an API. Complexity is the enemy of reliability. If you are building a serious product in 2026: Start with the raw SDK. Add abstractions only when you feel the pain. Not before. Are you Team LangChain or Team Raw Python? Let the war begin. 👇 #SoftwareEngineering #LangChain #AI #Python #DevOps #CleanCode #TechDebate

🚀 Built AutoRetry at Backend Reloaded Hackathon (2025) Many backend systems still use simple retry loops when jobs fail. This often causes retry storms and wastes system resources. During the Backend Reloaded Hackathon (15–21 Dec 2025), I built AutoRetry to handle failures more smartly. 🔹 Built using Motia 🔹 Uses step-based workflows instead of blind retries 🔹 Supports TypeScript, Python, and JavaScript 🔹 Decides when to retry, wait, stop, or escalate failures 🔗 GitHub: https://lnkd.in/dzFPFUmM Thanks to Kunal Kushwaha and the WeMakeDevs team for organizing a hackathon focused on real backend problems. Key takeaway: Good reliability comes from smart decisions, not more retries. #BackendReloaded #BackendEngineering #Reliability #DistributedSystems #Hackathon #OpenSource

AI-Assisted Code Modernization Recently, I needed to modernize a Panel app for busting bad colormap. I built it in 2020, and  froze it in time with Python 3.7 and Panel 0.x. This was deliberate to keep it functional, but after years, it caught up with me, in spite of some fixes last fall. The Binder deployment was also broken and I’d failed to fix on my own.Working in VSCode with Claude Opus 4.5 changed everything. In a short session, we upgraded to Python 3.12, fixed the deployment, updated an outdated widget pattern that Claude independently diagnosed, and quickly brought online some feature enhancements I had planned long ago, like adding two new colormap collections. The honest trade-off: In 2020, I learned Panel internals deeply through hard hands-on work and discussions in the Panel forums. In 2026, I got a working, modernized app in a fraction of the time, but with less low-level knowledge. For this project, that trade-off made complete sense; in fact, it was a game-changer, and that’s the path forward, for me. Read the full post on my blog at:https://lnkd.in/eWPWsCX8