Senior Platform Engineer

About Scalar
:

Building on a great API is one of the best feelings in software — every interface you want is right there, and it all just works. Building on a bad one is the opposite: out-of-date docs, no SDK in your language, no MCP server. We started Scalar to fix that.

We're a small team (around ten engineers and designers) who wear a lot of hats and obsess over our craft. We think every company deserves Stripe-level docs, SDKs, and MCP servers from day zero — not just the ones with a platform team to build them. Open‑source is in our DNA: our whole team contributes to and maintains repositories across the API ecosystem, including the standards everyone relies on, like OpenAPI.

We keep things lean and move without layers of process — engineers pick up problems, talk to each other, and ship. We're looking for owners with high agency: people who see what needs doing and do it. If you do your best work with a lot of autonomy and very little ego, you'll feel at home here.

You'll own the platform that every other Scalar product runs on — the services, infrastructure, and internal tooling behind Docs, SDKs, the API Client, and Agent Scalar. This is one of the most load‑bearing roles on the team: when you make the platform faster and more reliable, everyone ships faster.

Our infrastructure is real software, not config soup. We run a Type Script monorepo on GCP, and our infrastructure‑as‑code is Pulumi written in Type Script — so platform work here is engineering, not just YAML. You'll get deep into the existing system quickly and make it your own.

• Docs — the documentation platform:
  Markdown/MDX, OpenAPI and AsyncAPI references, two‑way Git sync, and AI/MCP chat, owned from backend to UI

• SDK generation — turning a customer’s OpenAPI document into type‑safe client libraries across Type Script, Python, Go, PHP, Java, and Ruby, with code samples and automated Git Hub sync

• The API Client — our open‑source, offline‑first Postman alternative, built on open standards and shipping across macOS, Windows, and Linux

• Agent Scalar — secure MCP servers and AI features for chatting with APIs, built on the Vercel AI SDK, the MCP SDK, and Vertex AI

• Our hybrid compute model:
  Cloud Run for most backend services, plus multi‑region Compute Engine managed instance groups for the stateful, high‑throughput ones (hosted doc serving, realtime, collaborative editing)

• Infrastructure‑as‑code in Pulumi (Type Script) — provisioning, secrets, networking, and the GCP services underneath everything

• The eventing and async backbone:
  Pub/Sub (with dead‑letter handling),
  Cloud Tasks
  , and Cloud Scheduler

• Object storage and artifact pipelines on GCS — cached docs builds, publish artifacts, image uploads, and CDN delivery

• The SDK generation pipeline — the service that turns OpenAPI documents into SDKs and keeps them synced to Git Hub

• Reliability, performance, and autoscaling as usage grows, across both Cloud Run and regional instance groups

• Internal tooling the whole team depends on: our in‑house CLI toolkit, per‑service deploy pipelines, Git Hub Actions CI, and our DB‑backed feature‑flag system

• Observability and operational health — Sentry, structured logging (pino → Better Stack), and the analytics path into Big Query

• You've owned production infrastructure at a startup or small team and liked the breadth of it

• You're comfortable being the person who understands how things actually work, end to end

• You can go deep on a Type Script‑heavy stack — and you like that our infra is written in Type Script too

• You've run real services on a cloud provider (we're on GCP) and understand the trade‑offs between serverless and long‑lived VMs

• You make decisions and communicate them clearly without waiting to be told what to do

• You'd rather collaborate to the right answer than win the argument, and you operate with a high level of kindness

• Depth in GCP specifically — Cloud Run, Compute Engine, Pub/Sub, Cloud SQL, GCS, Secret Manager, Artifact Registry

• Experience with
  Pulumi (or IaC in a real programming language rather than HCL)

• Multi‑region deployment, autoscaling, and rolling‑update experience

• A track record of building…