Teco

I build small, sharply-scoped scrapers.

Most scraping tools fail quietly. The target site changes, the tool keeps reporting success, and you find out only when your data is already wrong. So I test against live targets before shipping, and I publish what I actually measured — including the parts that failed.

For the YouTube subtitle scraper that meant running 60 real videos and comparing proxy types side by side. Datacenter proxies looked cheaper and twice as fast, but silently dropped the caption track on 3 videos out of 10 — no error, just missing data. That is why residential is the default.

Maintenance: I run a health check and review issues every week. If something of mine breaks, open an issue and I will tell you what I find, even when the answer is "this cannot be fixed right now".