Running Label Studio on my own self-hosted infrastructure (Proxmox homelab, not a third-party SaaS), with each client's project fully siloed. This is the actual annotation tooling behind Innerhouse Data, bounding boxes, classification, whatever the task calls for, configured and run in-house.
1
28
Self-hosted, built from scratch, dark blueprint theme with the tagline that says it all: "Your data never leaves the house." No cloud routing, no crowdsourced platform in between, just a small dedicated team and a client relationship you can actually reach by name.
0
12
Wrote a bash script to analyze how often Hubstaff (the time-tracking software gig platforms use) takes screenshots of your screen. Turns out the "random" intervals aren't random at all, there's a consistent upper bound and average gap. Small project, but it's the same instinct I bring to Innerhouse Data: don't just take a platform's word for how it works, go verify it yourself.
0
28
Nearly a year of annotation and model evaluation work on Outlier, consistent volume across multiple months. This is the ground-floor experience that led me to start Innerhouse Data: I saw firsthand how these platforms work (and where they fall short) from the inside.