Sobre este puesto de Platform Architect en Serverfarm
The next InCommand is a new solution, built on a new technology stack, designed from the ground up for the way modern data centers, including high-density and AI infrastructure, are operated. It will be delivered as a cloud service and be deployable in customers' own environments, built API-first, and developed with an agent-assisted engineering practice in which coding agents work from specifications under human review. Our own facilities are the first customer, so everything we build is proven in real, live operations before anyone else sees it.
At the heart of that product is its platform: one model of how a data center works, and the architecture that serves it fast to a live operations center and to AI working over years of history. We are looking for a hands-on Platform Architect who designs that foundation and writes the first code with AI-assisted tooling: the kind of architect whose decisions are in the repository, not only in the decision record. It is a rare chance to define a platform from a clean sheet, for an industry evolving faster than ever under exponentially growing demand for compute and AI, while standing on a decade of operational knowledge about what it has to do. You partner closely with Product Management, Integrations and Developer Platform.
Key Accountabilities
-
Build the foundation of the new platform yourself: the control plane and tenancy that everything else sits on, written with coding agents and the engineers who join you, and own the first code merged on the new platform.
-
Design and own the unified data model and its evolution, publish it with the public API as a specification others build against, and define the platform architecture around it (data stores, identity and multi-tenancy, API layer).
-
Own retrieval and query performance at telemetry scale: storage choices, query patterns for live consoles and for analytical and AI workloads, latency targets, tenant isolation at the data layer.
-
Choose the technology stack and platform standards, and recommend the cloud the platform starts on and how it stays portable.
-
Design how the platform integrates large language models and other AI services to deliver operational intelligence: insights, recommendations and natural-language access to operational data.
-
Define, in partnership with Integrations, the contract by which site data enters the platform.
-
Give technical direction to the engineers who join the new platform in its first months, and set the patterns the team builds on.
Required Skills
-
10+ years building back-end and platform systems, and in the last three years you have personally shipped a multi-tenant SaaS platform or control plane on Kubernetes, writing production code, as the technical decision-maker.
-
You have designed the domain data model for a complex operational system (assets, hierarchies, time series, events) and owned its evolution across releases, with the compatibility, versioning and migration decisions that came with it.
-
You have delivered retrieval performance at telemetry scale: chosen time-series and relational storage (PostgreSQL with TimescaleDB or equivalent, with a columnar engine such as ClickHouse for analytical and AI workloads), designed queries for live and analytical workloads over long histories, and met API latency targets.
-
You have built hands-on on a major public cloud (GCP, AWS or Azure) with Kubernetes and Helm, with judgment about when a managed service is the right call and when portability matters more.
-
You have built identity, tenancy and security by design into enterprise SaaS.
-
You have designed and published an API or data-model specification that other teams, and ideally external organisations, built against, and written architecture decisions that engineers worked from.
-
You have been a senior individual contributor on a platform's foundational production code.
Preferred Skills
-
Used agentic coding tools and directed and reviewed agent-written code at volume.
-
Knowledge of or experience with DCIM products: the data they manage and how data-center operators use them.
-
Designed platform ingress with Kafka-compatible streaming (Kafka, Redpanda or equivalent) and MQTT brokers (EMQX, HiveMQ or equivalent).
-
Instrumented a platform with OpenTelemetry for its own observability.
-
Integrated large language models into a platform: a model gateway, retrieval over operational data, tool interfaces an assistant can call.
-
Run a platform across public clouds, or shipped software both as a cloud service and in customer-hosted or air-gapped environments.
