Stop Engineering Your Prompts and Just Talk to the Thing
Most developers overcomplicate their AI coding workflow with elaborate setups and prompt rituals. The agents are good now. The bottleneck is you.
I help teams to build software by bridging Product, Design and Engineering.
I enjoy building software products in small, iterative cycles with clear quality gates. I work mostly with React,
TypeScript,
Node.js and
Swift for iOS apps. For backend work,
Python with
PostgreSQL and Docker.
Lately, I've been exploring Rust and ratatui for building small and performant terminal-based apps. Always learning and experimenting.
🏋 L0ad is a simple toolkit for Load Testing HTTP servers.
🖵 A small, fast and minimal system monitor readout that renders only essential sysinfo.
A scheduler server built with Node.js, TypeScript using DDD and Clean Architecture principles.
🗠 A tiny, minimal system monitor built to rendered as small herdr/tmux dashboard panes.
:incoming_envelope: Mail sending microservice built with Express.js, Docker and MongoDB
My shell and application configuration files, managed with GNU Stow.
Most developers overcomplicate their AI coding workflow with elaborate setups and prompt rituals. The agents are good now. The bottleneck is you.
Junior engineers ship faster than any generation before them. They also skip the struggle that builds intuition. That tradeoff is going to cost us.
A walkthrough of how I provisioned a Hetzner VPS with Terraform, deployed Dokploy, and self-hosted my entire automation infrastructure for the cost of a coffee per month.
The unglamorous realities of building production APIs around Large Language Models. Latency, cost, context windows, and why streaming isn't optional.
How I built a workflow that fetches news, generates social media posts with an LLM, and schedules them automatically. My content robot works harder than I do.
Email me if you think I'm the right fit and we should be working together on interesting projects.