Real lessons from shipping voice AI, high-traffic platforms, and AI products — from 7+ years as a solo founder.
The concrete optimizations that took opus from janky to silky-smooth on sub-$200 Android devices — FlatList tuning, JS thread management, Hermes, and the things that actually move the needle.
How I built manju's real-time image text extraction and translation system — the technical decisions behind the OCR pipeline, model selection, latency optimization, and handling text in complex image layouts.
Real lessons from 7 years building and scaling a voice AI platform in production — context windows, latency, edge cases, and why most voice demos fail in the real world.
Real techniques that move the needle in production LLM systems — structured outputs, chain-of-thought routing, few-shot selection, and why most prompt engineering advice is optimized for demos, not production.
I've been building production products with AI coding tools for over a year. Here's the honest breakdown of what vibe coding changes, what it doesn't, and how I actually use it.
After building both PWAs and native apps in production, here's the honest framework for choosing between them — capabilities, distribution, performance, and the cases where the conventional wisdom is wrong.
The specific optimizations that took yomimanga from a failing Core Web Vitals score to green across LCP, CLS, and INP — and why most performance guides miss the techniques that matter most at real scale.
The actual architecture behind yomimanga — how a solo founder built and scaled a manga platform to 5M+ daily requests without a devops team, and the decisions that made it possible.
MCP (Model Context Protocol) quietly changed how AI agents are built. Here's what it is, why it matters more than most people realize, and how it changes the way I build AI products.
Building powerx revealed every unsolved UX problem in self-custodial crypto wallets. Here's the honest breakdown of what makes Web3 wallet UX hard, what we built to fix it, and what the industry still gets wrong.
The solo founder / product engineer model has fundamentally changed. AI tools, global infrastructure, and distribution channels mean one person can do what a team of ten did in 2020. Here's what that looks like in practice.