The Blog
Raw thoughts on engineering, design, AI, and the art of building things that feel genuinely alive.
Why Interfaces Should Feel Alive — Not Just Look Good
True interface design goes beyond color palettes and type scales — it's physics. It's timing. It's the micro-moment when your finger lifts and a card springs back with just enough resistance.
The Quiet Revolution: How LLMs Are Changing What Engineers Actually Do
LLMs haven't replaced engineers. They've amplified the bottleneck. Here's what nobody is talking about: the shift from "writing code" to "describing systems".
Building APIs That Don't Apologize for Themselves
A well-designed API is invisible. A poorly designed one becomes a permanent fixture in your therapy sessions. Here's my framework for REST APIs that are actually a pleasure to consume.
Founding Kalki: The Part No One Puts on the Landing Page
Starting a creative agency at 19 looked glamorous from the outside. Inside it was spreadsheets at 2 AM, Figma arguments, and learning that "creative vision" doesn't pay server bills.
60fps or Nothing: GSAP Animation Without the Jank
Your animations shouldn't feel like they're apologizing. A complete guide to GPU compositing, will-change, and GSAP timelines that hold 60fps even on low-end Android devices.
The Portfolio Lie: Why Most Developer Portfolios Fail (And Mine Almost Did)
Most portfolios are a graveyard of todo apps and weather widgets. I rebuilt mine from scratch 4 times. Here's what I finally got right and what the data actually says about what clients look for.
RAG vs Fine-Tuning: When to Use Which (An Honest Answer)
After building 6 AI-powered products, the answer to "should I fine-tune or RAG?" is always the same: it depends — but here's a decision tree to actually make the call.
Glassmorphism Is Not Dead — You're Just Using It Wrong
Frosted glass had its moment and then got overused to death. But done right, blur + translucency creates depth that flat design simply cannot. Here's the correct pattern.
Python Async Is Not Magic — Here's What It Actually Does
I've seen junior devs sprinkle `async/await` everywhere like salt and wonder why their code is still slow. Event loops, coroutines, and tasks explained without the handwaving.
Pricing Your Work as a Freelance Engineer: Stop Hourly. Go Value.
Charging by the hour is a trap that caps your income and your freedom. How I moved to value-based pricing, lost two clients who didn't deserve me, and doubled my effective rate.
How I Cut My Web App's LCP from 4.8s to 0.9s (Real World, No Tricks)
Core Web Vitals aren't a vanity metric — they're a UX metric. A post-mortem on a real production site's performance disaster and the systematic fixes that actually moved the needle.
The 1-Year Review: What Building in Public Actually Did For My Career
After 12 months of sharing builds, failures, and lessons publicly — here's the real breakdown: followers, clients, lessons, and the things I'd never do again. No vanity metrics.
Building AI Products That Don't Feel Like Demos
How to move beyond the wrapper and build AI tools that solve actual user problems instead of just looking like the future.
Micro-Interactions in React: A Deep Dive into Framer Motion
The difference between a generic UI and a premium experience is measured in milliseconds. Here's how to build physics-based spatial transitions.
WebGL at 60fps: Optimizing Three.js for Production
3D experiences on the web are amazing until they melt your user's phone. A practical guide to draw calls, InstancedMesh, and geometry optimization.
The Myth (and Reality) of the Design Engineer
Why the industry is shifting from pure designers and pure devs to the people who can live in the bridge between them. The death of the traditional handoff.
Scaling WebSockets: The Architecture of Real-Time Apps
Building a collaborative canvas or live chat? Here is what happens when you move from 10 to 10,000 concurrent connections. Redis Pub/Sub, heartbeats, and binary payloads.
Why Local LLMs Are the Next Major Shift in Software Architecture
Cloud AI is great, but running models locally in the browser via WebGPU changes everything about privacy, offline capabilities, and compute costs.