Blog
Writing & Thoughts
Sharing what I learn about building software, AI, and product development.
The Most Expensive Skill I Have Is Knowing When to Quit
The longer you work on something, the more real it feels, and the more the sunk effort distorts the math. You don't evaluate the project anymore. You evaluate your relationship to the project. You defend it the way you'd defend a decision you've already made, because abandoning it now feels like admitting the past version of you was wrong. He was. That's fine. The past version of you was working with less information than you have today. Treating his conclusions as binding is the actual mistake, not the quitting.
Stop Talking to One AI. Start Running a Team of Them
Most AI tools give you a faster typist. One chatbot, one voice, one flattened answer. That stops working the moment your bottleneck shifts from getting answers to making better decisions. LiveCrew replaces the single-chatbot model with a crew of specialized agents you direct like a CEO. Four roles, real disagreement, structured meetings. Not a smarter assistant. A different shape of work.
I Solved 650 LeetCode Problems and It Didn't Make Me a Better Developer
Competitive programming taught me how to think. Production engineering taught me that thinking isn't enough. After 650 LeetCode problems and 150+ on Codeforces, I landed my first real client, and realized almost nothing I'd practiced for hundreds of hours mattered in production.