"I do MLOps, DevOps, and SRE. Essentially I break things, fix things, and then teach robots to break them faster."
| Role | What I Actually Do |
|---|---|
| 🤖 MLOps | Deploy models that were "99% accurate" in notebooks and watch them hallucinate in prod |
| 🔧 DevOps | Automate everything so I can be interrupted by alerts instead of humans |
| 🚨 SRE | Write postmortems at 3am. Current streak: too many |
resource "shilucloud" "identity" {
roles = ["MLOps", "DevOps", "SRE"]
on_call = true # always
sleep = false # see above
coffee = "critical" # lifecycle { prevent_destroy = true }
model_in_prod = "it depends"
uptime_sla = "99.9%" # personal SLA: 60%
}
08:00 — Write Terraform. Feel like a god.
09:30 — terraform apply. Feel like a fool.
11:00 — ML model is "drifting." So am I.
13:00 — Lunch. Alert fires. Lunch cancelled.
15:00 — Postmortem: "human error." (it was me)
17:00 — Fix it with a pipeline. Ship it. Go home.
02:00 — PagerDuty says otherwise.
- ✅ Secure Terraform pipelines that haven't nuked prod... yet
- ✅ LocalStack: break things locally so prod can break in new creative ways
- ✅ CI/CD pipelines more reliable than my sleep schedule
- ✅ Deployed a model. It was 99% accurate. In the notebook.
- ✅ SLOs I set and SLAs that set me
█████████░░░░░░░░░░░ 42% remaining
(Please stop deploying on Fridays. I am begging you.)
"Ship fast. Break things. Write the postmortem. Repeat."
⭐ If my infra hasn't paged you yet, give it time.