National Oilwell Varco — Go Microservices at the Drilling Edge

01 The Challenge

National Oilwell Varco — one of the world’s largest oilfield equipment and services companies — runs a drilling-data platform whose services live in two worlds at once: on edge devices at drill sites and in the AWS cloud. Data captured at the rig has to flow reliably up to the cloud, and platform capabilities have to work consistently in both places.

The platform needed new production services — asset management, company, and identity/authorization — plus streaming APIs, and it needed to shed hardcoded assumptions that tied the API to a single company. On top of the engineering work, the platform’s security model was undocumented, and the team’s delivery process had room to tighten. Leopard Data was brought in to work all three fronts.

02 The Approach

Leopard Data built Go microservices with gRPC and Protocol Buffers streaming APIs and generated gateways, backed by PostgreSQL with goose-managed schema migrations. Edge-to-cloud data synchronization flowed NATS → MQTT (Mosquitto) → Kafka, moving drilling data from rig-side devices into the cloud pipeline. We implemented streaming endpoints such as wells-by-operator, classification endpoints with edge sync, and application-scoped roles and permissions endpoints — and made the platform API company-neutral, removing hardcoded assumptions so it could serve multiple companies.

On security, we fully analyzed the platform’s existing security model and wrote the security document and sequence diagrams the team then used as its reference. We designed SAML and OpenID Connect support with Okta, including token-expiry handling, and delivered working SAML applications for dev and test alongside the security admins.

On process, Leopard Data took over the scrum-master role so the product owner could focus on the product: running stand-ups, retros, and demos; instituting a daily pull-request review meeting; restructuring Jira into epics, features, and stories; and writing the team-standards document and story-writing standard the team adopted. We also mentored the team on microservices architecture in recurring sessions.

03 The Delivery

Endpoint sets shipped to production with unit tests and a clean QA handoff — streaming wells-by-operator, classification with edge synchronization, and application-scoped roles/permissions all went from design to running services. The company-neutral API work landed without disrupting the existing platform consumers.

The security deliverables were working artifacts, not shelfware: the security document and sequence diagrams became the team’s shared reference, and the SAML applications for dev and test were validated end-to-end with the client’s security administrators. Meanwhile the process changes took hold in the daily rhythm — shorter stand-ups, PRs reviewed every day instead of piling up, and stories written to a standard everyone could work from.

05 Tech Stack