devan raut
platform engineer — bengaluru
i keep the boring parts of production boring: kubernetes clusters, delivery pipelines, and the tooling around them. most of what i know i learned from incidents; most of what i write exists so the next one is shorter.
open to platform work from january
about
i've spent ten years on the plumbing side of software — the clusters, queues, and build systems everything else quietly depends on. i like this layer because it rewards patience: the wins are invisible and the mistakes are educational.
day to day that means running kubernetes for teams who'd rather not think about it, maintaining two open-source tools that outgrew their original weekend, and writing things down — runbooks, postmortems, long design docs nobody asked for and everybody ends up reading.
i work remote-first from bengaluru, keep a small bare-metal lab under the desk, and hold the possibly unpopular opinion that most incidents are documentation failures wearing a trench coat.
- location bengaluru, in (utc+5:30)
- focus infra & distributed systems
- stack go · rust · kubernetes · nix
- writing runbooks, postmortems, rfcs
services
platform engineering
clusters, delivery pipelines, and a paved road between a commit and production that nobody has to think about.
reliability work
slos that mean something, alerts that page for a reason, and postmortems people actually read.
infrastructure reviews
a second pair of eyes on the terraform before it becomes a load-bearing mistake.
internal tooling
small sharp cli tools for the workflow your team repeats twenty times a day.
sound like your backlog?
start a conversationwork
| year | project | stack | role | open |
|---|---|---|---|---|
| 2026 | relay | go · cobra · sqlite | author & maintainer | |
 a single-binary cli that tails, filters, and diffs structured logs across clusters — because grep deserved a promotion, not a replacement. | ||||
| 2025 | fleetd | go · kubebuilder · redfish | design & implementation | |
 a kubernetes operator that manages bare-metal node pools like cattle — provisioning, draining, and reimaging without a human in the loop. | ||||
| 2024 | quorum | rust · tokio · raft | maintainer | |
| 2025 | edge lab | nixos · k3s · arm | everything, unfortunately | |
| 2024 | paved road | buildkite · terraform · go | lead engineer | |
| 2023+ | protocol notes | prose · rfcs · patience | author | |
resume
experience
2022—present
senior platform engineer — datum works
2019—2022
site reliability engineer — arcline
2016—2019
backend engineer — sequent labs
education
2012—2016
b.e., computer science — bms college of engineering
references
the etcd migration nobody wanted to own. devan owned it, wrote the plan, and the plan was boring — which is the highest compliment i have for infrastructure work.
his runbooks read like they were written by someone who has actually been paged at 3am and would prefer you weren't.
contact
no funnel, no calendar link — a plain email works best. a few lines about the system, the problem, and where you'd like it to end up.
or use the form
got it — thanks for writing. i'll reply within a day or two.