Peter Czimmermann

A whole engineering team working with coding agents every day

Owned the engineering AI tooling programme end to end. Ran the evaluation that selected Claude Code over the alternatives, own its budget and spend controls, wrote the usage and data-handling guidance for what code and data may be sent to AI tools, and tracked adoption and outcomes for leadership. Built a 20-week plan to grow the team’s AI capability and a recurring company-wide show-and-tell where people demonstrate what they build with AI.

Designed and delivered an in-person, hands-on workshop across SensorUp’s Vancouver, Calgary and Toronto offices in April 2026. Every engineer who attended set up the tooling and ran real tickets through it in the room. The aim was as much confidence as skill, so that the change in the profession felt learnable rather than threatening.

Built the internal Claude Code plugin tailored to SensorUp’s stack (a GraphQL federation skill, a Terraform reviewer, workspace setup and release commands, review-gate agents, MCP servers, and a Claude Desktop bundle for non-engineers) and the process around it: decisions written as architecture decision records with an automated lifecycle, short challenge sessions to settle open questions, and reviews routed by code ownership so that agent-assisted work merges safely without waiting on the architect. This is how the full-stack rebuild below reached partner testing in six months.

• Governance: tool evaluation and selection, budget and spend controls, usage and data-handling guidance, adoption reporting • Enablement: three-city hands-on workshop, 20-week capability plan, company-wide AI show-and-tell • Tooling: Claude Code plugin authoring (skills, agents, hooks, MCP servers), review-gate agents • Process: architecture decision records as the source of truth for people and agents, ownership-routed reviews

SensorUp builds emissions and decarbonisation software for energy and heavy industry, backed by Climate Investment

SensorUp brings field alerts and sensor data into one operating record, where AI agents sort the signals and propose the next step while people keep the decision. Its products cover methane, carbon-credit and greenhouse-gas reporting, and its current platform is SensorUp Studio, which lets customers tailor those workflows without new software. The company is backed by Climate Investment, the decarbonisation investor founded by the Oil and Gas Climate Initiative.

Lead the Architecture and Platform teams and the IT, security and compliance function, six people in total. Own the architecture decision log and specification process that every squad builds against, review cross-cutting designs, and resolve pattern conflicts between squads. Grew two tech leads into acting architects for their own squads. Own the 2026–2027 annual engineering initiative to turn agentic development into a repeatable product-delivery pipeline.

In 2026, led the design and rebuild of SensorUp’s full stack as SensorUp Studio, a multitenant platform, from prototype to the first partner testing round in six months. Wrote the architecture down as decision records covering tenancy, isolation, identity, composition, authorization and cross-module links. Designed and shipped the identity and access layer end to end: sign-in, tenants, invitations, roles, and authorization checked on every write.

From 2021 to 2025, owned the architecture of a federated, serverless estate of about twenty domain services and carried every fleet-wide change through it as coordinated, backward-compatible rollouts: the standard deployment contract, the security-hardened infrastructure migration, runtime and framework upgrades, and the engineering controls for SOC 2. Built the deployment tooling the team ships through and the cross-account data pipelines that enterprise energy customers run on. Designed the configurable asset model behind the methane and carbon products. Rebuilt the release pipeline so each product stream deploys independently, retiring a separate production deployment repository and taking operational load off other teams. Started the engineering knowledge base, and covered part of the VP Engineering role for several months in 2025.

• Architecture: Apollo GraphQL federation, CASL authorization, central auth service, event-driven services on AWS Lambda, DynamoDB single-table design, change streams exposed as queryable datasets • Delivery: GitHub Actions, Spacelift (Terraform / OpenTofu), AWS CodeArtifact, ECR, per-stream deployments, EU region enablement • Data and integration: data catalog and task-runner platform, cross-account S3 / SFTP / DataSync movement, Dagster, OGC SensorThings • Security and compliance: SOC 2 preparation, trust-policy hardening, backup and restore testing

Joined a small team and designed the AWS infrastructure blueprints that every customer deployment is built from, the Explorer web application’s API, and the SensorHub Android app that collects field sensor data. Introduced Apollo GraphQL federation as the integration backbone across a growing set of services, with JWT validation and user and permission propagation to every subgraph.

• Infrastructure as code: Terraform modules, Serverless Framework, per-customer deployment configuration, automated DNS and certificate bootstrap • Identity: Cognito user pools, IoT Core policies, CASL authorization propagation • Services and mobile: Node.js Lambda subgraphs, API Gateway, DynamoDB, geofence rules engine; Android (Kotlin / Java) GPS, Fitbit and raw-sensor telemetry

Defined and implemented the cloud architecture and software delivery platform for Kantar’s “new ways of working” programme on AWS, then worked inside each team in turn to drive adoption and adjust the tooling and documentation to what they needed. The result was a working continuous deployment model across the teams.

• Infrastructure as code with serverless.com and CloudFormation; build, test and deployment pipelines on CodePipeline and CodeBuild; ECS and Docker for stateful services, Lambda (Python), SNS and API Gateway for stateless ones

After the sale and merger, led the team that bootstrapped a microservices architecture to integrate the existing monoliths and expose the core systems through RESTful APIs, then led the move from a hosted data centre into AWS ECS, introducing Docker and cloud practices to the team.

• Scala and Play microservices with Couchbase; Docker, ECS and CloudFormation; Jenkins, Puppet and behaviour-driven testing

Led teams delivering panel management software operated as SaaS, in Hungary until 2007 and in Vancouver from then on. Introduced continuous integration, behaviour-driven testing and a fully automated build and deployment system to the engineering organisation.

• LAMP stack web applications; Behat, Selenium and PHPUnit; GNU make and RPM-based build and deployment

Built the original survey sampling engine and web interface in GMI’s start-up years, then a high-volume mail-merge engine for survey invitations that remained in use for over a decade.