· 4 min read
Kafka Double Payment Bug
How to prevent duplicate payment processing when Kafka workers crash using idempotency patterns
Writing about tech, leadership, and engineering practices
Sharing insights from my journey as a Lead Engineer, covering topics from technical architecture to team leadership.
· 5 min read
Post-Mortem: NPM Is Compromised
What we learned from the largest supply chain attack in JavaScript history
· 4 min read
The Developer's Five Rings
How Miyamoto Musashi's ancient strategy guide applies to modern software development mastery
· 4 min read
The Internet's Postal Service
How BGP routes your data across the world - and why it sometimes delivers to the wrong address
· 4 min read
IPL Ticket Wars Survival Guide
How to build a ticket booking system that won't crash when millions of cricket fans attack your servers
· 4 min read
IndexedDB Caching Made Simple
Learn how to implement robust client-side caching with IndexedDB for offline-first web applications
· 4 min
Smart Rate Limiting Strategies
Beyond basic rate limiting - adaptive techniques to stop bots while keeping real users happy
· 4 min
Real-Time Tracking Without the Drama
How to handle 100k GPS updates without melting your servers or draining every phone battery
· 4 min read
When Leaderboards Kill Games
How a simple SELECT query can bring down a 10M player mobile game and the Redis solution that prevents it
· 4 min read
Is Chain-of-Thought Reasoning Real?
Recent research reveals chain-of-thought AI reasoning might be sophisticated pattern matching, not genuine thinking
· 4 min read
GPU Costs Melting Your Budget
How semantic caching can save your AI chatbot from expensive GPU waste
· 4 min read
GPT-5 Finally Arrives
OpenAI's latest model combines reasoning with speed, but does it live up to the hype?
· 5 min
Understanding Pods in Kubernetes
A detailed look into how Kubernetes Pods work, how to create them, and why they're the foundation of all workloads.
· 3 min
Cognitive Load Theory: How Less Thinking Helps You Think Better
A deep dive into Cognitive Load Theory — why it’s a game-changer for learning, memory, and decision-making.
· 9 min
Vector Dimensions: Handle with Care
Understanding the role of vector dimensions in embeddings — what they mean, why they matter, and how to choose wisely.
· 8 min
Embeddings Demystified: Math, Meaning & Machines
A clear, human-style walkthrough of what embeddings really are — how they’re made, why they work, and what they power.
· 9 min
Synonymic Query Expansion for Smarter Search
A deep dive into how synonym-based query expansion improves search results in LLM-powered apps and enterprise tools. Real-world cases, design patterns, and tradeoffs.
· 9 min
Chain-of-Thought Prompting: Teaching LLMs to Think Out Loud
Part three in my prompt engineering series — this time we explore chain-of-thought prompting, where structured reasoning turns LLMs from guessers into guided problem solvers.
· 7 min
Few-Shot Prompting: The Middle Ground Between Effort and Accuracy
Part two in my prompt engineering series. In this post, I dive into few-shot prompting — when a little context goes a long way.
· 7 min
Zero-Shot Prompting: The Cleanest Trick in Prompt Engineering
The first post in a series on prompt engineering — this one explores zero-shot prompting: the cleanest way to get results from LLMs without examples, training, or fine-tuning.
· 6 min
Why SMART Goals Actually Work (When Done Right)
Not just another acronym. SMART goals can be your unfair advantage if you know how to wield them. Here’s how I make them stick.
· 11 min
FastAPI to ECS the Smart Way: Load Balanced & Custom Branded
From FastAPI app to scalable ECS Fargate deployment — all load balanced and wired to your custom domain.
· 10 min
Why React Virtualized Matters: Rendering Smart, Not Hard
A deep dive into React Virtualized — how it works, why it saves performance, and when to use it.
· 12 min
Context Without Complexity: LangChain’s In-Memory Superpower
Lightweight, fast, and surprisingly powerful — how LangChain’s in-memory store helped us manage chatbot context during a hackathon sprint.
· 7 min
Dependency Inversion Principle in C#: Flipping Dependencies for Cleaner Architecture
A practical guide to the Dependency Inversion Principle in C#—what it means, why it matters, and how to apply it without overengineering your code.
· 6 min
Interface Segregation Principle in C#: Build Interfaces People Actually Want to Use
A developer-friendly look at the Interface Segregation Principle in C#, showing how to design clean, focused interfaces without forcing classes to implement what they don't need.
· 7 min
Liskov Substitution Principle in C#: Subclass Without Surprises
A practical guide to the Liskov Substitution Principle in C# with real-world examples, helping you avoid sneaky inheritance bugs and unexpected side effects.
· 7 min
Open/Closed Principle in C#: Making Your Code Flexible Without Breaking Stuff
A practical guide to the Open/Closed Principle in C#, using real-world examples to show how to extend functionality without modifying existing code.
· 6 min
Single Responsibility Principle in C#: Writing Classes That Do One Thing Well
Understanding SRP in C# with relatable examples, real-world refactoring, and practical insights for cleaner, more maintainable code.
· 10 min
How LLMs Understand Images: The Secret Behind AI That Sees
A deep dive into how AI models like GPT-4V and CLIP process images, bridging the gap between vision and language with Transformers and multimodal learning.
· 12 min
LLMs & Transformers: A Beginner’s Guide for Software Engineers
A practical, easy-to-understand guide on Large Language Models (LLMs) and Transformers—how they work, why they matter, and how they impact modern software development.
· 5 min
Jotai and Zustand: Modern Alternatives to Redux
Exploring how Jotai and Zustand offer lightweight, efficient, and developer-friendly state management solutions compared to Redux.
· 9 min
Managing Backend State with React Query: Why Redux Isn’t Always the Answer
Explore how React Query simplifies backend state management in React applications, reducing boilerplate and improving your app’s maintainability—without Redux.
· 8 min
Gemma 3: DeepMind's Leap in Multimodal AI
A deeper look at DeepMind’s Gemma 3 — a powerful, multimodal AI model that handles long contexts, over 140 languages, and runs efficiently on a single GPU. What makes it different, and why should you care?
· 6 min
Taming Circular Dependencies with Kahn’s Algorithm
A quick, practical look at how Kahn’s Algorithm can prevent runaway cycles by neatly ordering tasks and modules. Learn how to spot circular dependencies, fix them, and keep everything running smoothly.