Christophe Delattre

Dec 2025 – Present · Munich, Germany

Leading the evolution of the AUD (Amadeus Unified Descriptor) network-automation framework — which I originally designed and built — into a generic, open-source product: re-architecting its model to fit any use case beyond internal needs, while keeping security and governance under control.

• Re-designing the AUD data model and architecture to be generic and use-case-agnostic, fitting any network-design scenario rather than only Amadeus ones.
• Driving cross-organisation requirement gathering and validating the new model against those requirements.
• Project planning to align the open-source direction with internal requirements, security and governance.
• Proposing multiple architecture and design options, with graphical tooling to help users design networks against the new model.
• Authoring the high-level design and low-level specifications, with implementation under way and CI on GitHub and GitHub Actions.
• Choosing PostgreSQL as the backend and defining its operational procedures, SLIs, SLOs and observability, backed by performance studies across the target use cases.

• PostgreSQL
• GitHub
• GitHub Actions
• SLI/SLO/SLA
• Observability
• System Design
• Open Source
• Governance

Jan 2024 – Dec 2025 · Munich, Germany

Led the reliability-focused integration of a newly acquired hotel-technology product into the group’s engineering standards — migrating its bespoke tooling, processes and platform to ensure a smooth transition with uninterrupted service continuity.

• Defined SLOs deliberately tighter than the contractual SLAs — 99.95% availability and 300 ms response time against 99.9% and 500 ms — to keep a margin before customer commitments were at risk.
• Introduced RCA and post-mortems where none existed, chaired the reviews myself to set the standard, tracked fixes to closure and blocked releases until they landed, under a "no incident happens twice" rule.
• Integrated end-to-end observability — metrics and monitoring with Prometheus, alerting with Alertmanager, and centralised logging with Splunk — cutting log latency from 15 minutes, with occasional loss, to near-real-time so SLO attainment could actually be seen.
• Built CI/CD pipelines with Jenkins and GitHub Actions to produce consistent, repeatable deliveries — moving from monthly releases, to weekly releases.
• Standardised deployments using Helm umbrella charts and ArgoCD (GitOps), and evaluated a replacement for the legacy software-promotion process.
• Migrated ~40 microservices from virtual machines to Kubernetes across four production clusters — on-prem and cloud, each split PCI and non-PCI — plus two test clusters, and planned major platform evolutions including cluster creation and migration.
• Integrated the product with the group’s change-management process and kept the PCI environments compliant — regular security patching of infrastructure and applications, plus support to development teams on their Black Duck and SonarQube scans and security reporting.
• Reworked platform access controls — more visibility for developers, tighter control for SREs — and delivered self-service, ephemeral environments for developer integration testing.
• Defined the RACI matrix that had never existed — splitting ownership between platform, SRE and application teams — then coached the acquired teams, who had no prior SRE practice, documenting and spreading the standards behind each new process rather than imposing it.
• Worked within a 10-person team across three regions (Americas, Europe, Asia) on a follow-the-sun weekly on-call rota, and helped recruit two engineers in India.

• SLI/SLO/SLA
• Kubernetes
• Helm
• ArgoCD
• GitOps
• Jenkins
• GitHub Actions
• Prometheus
• Alertmanager
• Zabbix
• Datadog
• Splunk
• CI/CD
• PCI
• SonarQube
• Black Duck

Dec 2019 – Dec 2023 · Munich, Germany

Technical lead within the Network Automation team, which provides and operates the network infrastructure behind all the group’s products. Helped shift the team from manual, change-heavy network operations toward a self-service, software-defined networking (SDN) model, building the frameworks that let users declare their network needs and have devices configured automatically.

• Technical lead of the team: set technical direction and helped grow it from 2 to 14 engineers by taking part in recruitment and ramp-up.
• Designed and built AUD from a blank page — an OpenAPI-modelled data structure and orchestration layer (Azure Functions, multi-layer caching) that became the backbone for firewall, NSG, OVN and security-dashboard automation, and is now used by every engineering team in the group.
• Took firewall changes from manual, device-by-device configuration to a fully automated process spanning the successive cloud security layers, driven by declared intent instead of hand-written rules.
• Reduced change risk by automating the change-management process for the AUD applications (Viewer, API, Orchestrator) through ServiceNow, replacing manual approvals with auditable workflows.
• Established reliability guardrails across all framework Python services: code-quality standards, test strategies, KPIs, monitoring and alerting, and patch management.
• Authored operational runbooks and incident/problem-management procedures so operations teams can run the AUD framework and respond to incidents independently.
• Delivered DNS automation (record creation) on-prem via Ansible/AWX to a shared standard, then handed it over to the DNS team as a self-service capability.
• Achieved PCI compliance for the framework and delivered the CCB Approval Viewer giving change-review boards real-time visibility into technical-request progress.

• Terraform
• Ansible
• AWX
• Azure
• Azure Functions
• Jenkins
• Bitbucket
• ServiceNow
• OpenAPI
• Python
• SDN
• PCI

Sep 2017 – Dec 2019 · Munich, Germany

Owned the platforms and delivery processes behind the Airline Data unit’s analytics and machine-learning products — running the infrastructure and reliability side of data science rather than the modelling.

• Stream owner for the DSC (Data Science & Consulting) cluster: architecture, capacity planning and operations for a virtual team of 5 across 3 teams.
• Built full CI/CD for cluster deployment and lifecycle management (Jenkins, Azure DevOps) with Prometheus monitoring integration and managed releases.
• Stream owner for the Data Science Platform (DSP) v1: defined and automated it on OpenStack, including data-flow orchestration.
• Led the evaluation and rollout of notebook platforms (JupyterHub, Zeppelin), proving them out on Kubernetes/OpenShift.
• Industrialised delivery: standardised delivery pipelines and development workflows; ran Hadoop/AWS EMR clusters for performance-testing comparison against on-prem and a GPU PoC for security anomaly detection.

• Kubernetes
• OpenShift
• OpenStack
• Azure DevOps
• AWS EMR
• Hadoop
• CI/CD
• JupyterHub
• Python

May 2015 – Sep 2017 · Boston, USA & Munich, Germany

Responsible for continuous service delivery of the InterContinental Hotels Group platform — a brand-new technology stack in new datacenters built to meet challenging SLAs while keeping maintenance cost down.

• Owned service continuity for the first Amadeus service sold with a 99.999% availability and 50 ms response-time SLA — corporate SLOs did not cover that tier, so we defined new ones and backed them with near-real-time alerting on SLA breach.
• Ran the production estate behind those targets: 20+ OpenShift clusters (10 in production across 2 regions × 5 availability zones), 4 Couchbase clusters replicated with XDCR and 2 Oracle clusters on Data Guard — all sized from the contractual SLAs, traffic forecasts and application footprint.
• Enforced an error-budget policy in the early days — two production problems in a row blocked the next release until a clean run was evidenced — then took the change-failure rate close to zero with a fully automated Jenkins CI promotion chain catching failures upfront.
• Built infrastructure-as-code from the ground up — Infrastructure-as-a-Service on OpenStack/VMware, bootstrapped with Heat, Terraform, cloud-init, Puppet and Ansible, with applications deployed on Red Hat OpenShift.
• Designed and validated functional and non-functional monitoring and alerting: OpenShift and Oracle monitoring, SoapUI/Jenkins functional monitoring and a fully automated Nagios setup.
• Setup a follow-the-sun on-call rota across Boston, Munich and Sydney (12 engineers, weekly shifts, everyone in the loop): a single customer-impacting major incident over the period, all others auto-detected and resolved within a minute (MTTR), each with RCA, post-mortem and auto-remediation until the permanent fix shipped.

• SLI/SLO/SLA
• OpenShift
• OpenStack
• VMware
• Terraform
• Heat
• Puppet
• Ansible
• Jenkins
• Nagios
• Couchbase XDCR
• Oracle Data Guard

Apr 2013 – May 2015 · Munich, Germany

Lead developer of an operational tool suite that helps front-line teams manage incidents efficiently and reduce recovery time.

• Cut incident recovery time with fast, global and secure automated recovery actions, and reworked the architecture to meet SLAs — a multi-datacenter Couchbase design with a lightweight front-end to reduce latency for remote sites (Sydney, Miami).
• Introduced software-quality and release-management practices: versioning-to-CI release process, code coverage, static-analysis gates, and early Vagrant/Docker test environments.

• Java (J2EE)
• Couchbase
• Jenkins
• Git
• Docker
• Vagrant

Jan 2011 – Apr 2013 · Sophia-Antipolis, France

Middleware products owner for the Orange portal search-engine backend. Coordinated a team of 3 building a multithreaded inter-process communication framework and multi-datacenter caching layer — improving performance, robustness and response-time SLA compliance.

• ElasticSearch
• Middleware
• Caching
• RDF/XML

Jan 2008 – Jan 2011 · Carros, France

Embedded and client software for airfield and public-lighting monitoring and control systems: airfield lighting control, a control-tower monitoring client, and a public-lighting monitoring project run end to end for customer EDF.

• Embedded C/C++
• Rich client
• Monitoring