Senior Site Reliability Engineer (SRE)

Etienne Dallard

Building reliable cloud infrastructure and distributed systems at scale.

10 years of experience across Linux infrastructure, automation, and reliability, with production experience on Azure, OpenStack, Kubernetes, and Terraform.

I design and operate infrastructure and distributed systems from architecture through production, with a focus on observability, security, backup/recovery, and operability.

  • 10 yearsLinux · DevOps · SRE
  • ~1 PBDistributed systems · Backup · Recovery
  • Azure · OpenStack · KubernetesTerraform · Python · Go · GitLab CI/CD
Contact meDownload résumé (PDF)

Profile

With 10 years of experience across Linux infrastructure, DevOps, and SRE, my background spans systems and data center infrastructure administration through the design, automation, and operation of cloud infrastructure and distributed systems.

I have production experience with Azure and OpenStack, alongside Kubernetes, Terraform, Python, and Go. My main areas of focus are reliability, automation, observability, security, and recovery mechanisms.

I particularly enjoy problems that require understanding an existing system, evaluating several approaches, building a proof of concept, and making architecture decisions that balance technical, operational, and cost constraints.

My approach remains hands-on, from design through operations, with particular attention to operability, reliability, and the continuous improvement of production systems.

The case studies presented here illustrate this approach through the architectures chosen, technical decisions, trade-offs, and outcomes.

Professional Experience

Site Reliability Engineer (SRE)

Ubisoft Montréal, Montreal, Canada - hybrid

Dec 2024 – present

Perforce Backup System

Distributed Infrastructure · Restic · ~200 instances · ~1.3 PB

  • Compared Kopia and Restic on three representative production projects; restore times, rather than the backup window alone, drove the recommendation to adopt Restic, approved by leadership.
  • Perforce is the version control system used to manage large game-development assets. Designed distributed backup and recovery infrastructure for ~200 Perforce instances (~1.3 PB of source data, ~940 TB backed up), with per-instance scheduling and execution and a separate Restic repository and Azure container per instance to limit the impact of a repository failure.
  • Chef generates Resticprofile configurations and systemd timers from declared configuration; checks of real backup paths, consecutive-failure alerts, and Vault access via Azure Arc Managed Identity support operations without persistent Azure credentials on servers.
  • Led the production rollout and handover to the Ops team with runbooks and a restore guide; the solution reduced backup costs by about 87%, or roughly US$1M/year.

Perforce Backup Replication

Azure Functions · Python · ~940 TB · Disaster Recovery

  • Designed an independent copy of ~940 TB of backups across two Azure region pairs (four regions), with monthly delta synchronization and Cool/Archive tiering.
  • Durable Functions persists orchestration state for cycles that can last several days; bounded AzCopy concurrency, sharding, and checkpoints enable partial recovery. Locked repositories are requeued.
  • Managed Identity authenticates the runtime and OIDC the CI/CD pipeline; Terraform and GitLab CI/CD deliver the infrastructure, backed by unit tests and Azurite scenarios.
  • Weighed GRS (less software to operate) against a lower-cost independent copy for this scope, accepting responsibility for testing, monitoring, and recovering a custom pipeline.

Disaster Recovery Workflow

GitLab CI/CD · Azure Blob Storage · Disaster Recovery

  • Implemented a GitLab CI/CD workflow that prepares an instance for restore from replicated storage: rehydrates Archive data when required, transfers the repository to dedicated storage, and restores the Perforce instance with Restic.

db-backup Azure Migration

Azure Blob Storage · Kubernetes · AKS

  • Migrated db-backup storage from OpenStack to Azure Blob Storage, then moved the Kubernetes platform to AKS through Ubisoft's internal Kubernetes infrastructure, adapting the API and Kubernetes components.

Working Method

Claude Code · Architecture · Acceptance Criteria · Targeted Reviews

  • For Perforce Backup System and Replication, I defined the architecture, constraints, and acceptance criteria, then used Claude Code as a coding agent for implementation, testing, and documentation, with targeted reviews and validation before production rollout.

DevOps & Infrastructure Engineer

Ubisoft Montréal, Montreal, Canada - hybrid

Nov 2020 – Nov 2024

db-backup

Python/Django · Go · Ansible · Kubernetes · Backup Platform

  • Contributed to the development and production rollout of db-backup, then assumed primary ownership of the service. The centralized platform covered ~190 Redis/MySQL servers, replaced rsync/NAS and NetBackup, and delivered more than $200K in annual cost savings.
  • The Django REST API centralized backup requests and service state; Kubernetes CronJobs scheduled runs, while Ansible playbooks performed the MySQL and Redis operations. The Go CLI used the same API contract; Vault handled secrets, and metrics and dashboards made backup status visible.
  • Maintained and evolved the API, CLI, playbooks, and Kubernetes components; automated restore-then-backup tests in GitLab CI/CD on real instances, including weekly validation with production data on OpenStack VMs provisioned with Terraform.

Internal Services

Kubernetes · OpenStack · Prometheus/Grafana · NCSA · EMEA · APAC

  • Resolved more than 300 operational incidents related to db-backup (backup failures, credential rotation, storage capacity) and infrastructure services, covering the NCSA, EMEA, and APAC regions.
  • Operated Kubernetes and OpenStack environments and contributed to infrastructure changes.
  • Contributed to the evolution of monitoring and alerting with Prometheus, Alertmanager, Grafana, and Zabbix.

OS Modernisation CentOS 7 → Rocky 9

Linux · Migration · 3 regions

  • Migrated all Linux jump servers (8 servers, 3 zones: NCSA, EMEA, APAC) and legacy MySQL production servers from CentOS 7 to Rocky 9, with live IP transfers and no service interruption.

Incident Remediation Automation

StackStorm · R&D · Event-driven automation

  • Led an R&D evaluation of incident remediation automation tools, selected StackStorm, built a proof of concept, and presented a demonstration to management.
  • Another team subsequently developed the proof of concept into a production service, which remains in use by the Command Center.

Linux Systems & Infrastructure Administrator

Ubisoft Montréal, Montreal, Canada - hybrid

Mar 2017 – Nov 2020

Game Servers & Internal Services

~300 servers · 3 regions · 21 production incidents

  • Administered and automated several hundred Linux servers (CentOS/RHEL) via Chef and Cfengine on OpenStack/VMware VMs and bare metal, across three regions (NCSA, APAC, EMEA).
  • Diagnosed and resolved production incidents on online game backend services (Redis, MySQL, MongoDB, Couchbase, PowerDNS, Varnish, JumpServer), including CCU degradation investigations on live titles such as Watch Dogs, Rocksmith, AirMech, Just Dance, and Assassin's Creed.
  • Handled security and network operations (SSL/TLS certificates, DNS/GeoDNS/PowerDNS with cross-datacenter migrations, load balancers), global on-call rotations, and SVN access for development teams.
  • Implemented monitoring with Zabbix; delivered Linux and application migrations, and contributed to server provisioning and decommissioning (~37 operations).

Knowledge Transfer & Mentoring

Onboarding · Knowledge sharing · Hiring interviews

  • Onboarded new team members, led knowledge transfer, and participated in technical hiring interviews.

Linux Systems Administrator

Mikros Image Canada, Montreal, Canada

Mar 2016 - Mar 2017

Render farm & infrastructure datacenter

~1,300 nodes administered · 40 bare-metal servers

  • Administered a Linux render farm of approximately 1,300 nodes and deployed 40 bare-metal servers in the datacenter for the studio's 3D/2D rendering workloads, covering physical cabling, Linux installation, network configuration, and production rollout.
  • Administered CentOS 6 and RHEL servers hosting critical services including LDAP, FreePBX, Puppet, PXE imaging, and off-site backups.
  • Managed Hitachi NAS storage, including data migrations, snapshots, and performance troubleshooting, as well as Extreme Networks infrastructure installation and upgrades.
  • Provided Tier 1-3 technical support across Linux, Windows, and macOS; trained users on Fedora 20/XFCE; and managed the site's Centaur access-control system.

Systems & Network Administrator

Groupe Rodeus, Montreal, Canada

2013 – 2016

Windows & Client Network Administration

Windows Server 2008/2012 · Active Directory · DHCP, NAT, DMZ

  • Administered Windows Server 2008/2012, Active Directory, Exchange, and VMware ESXi 5.5 virtualization.
  • On-site and remote work: installations, migrations, networking (DHCP, NAT, DMZ), and L1/L2 support.

What teams can expect from me

I take ownership of complex work without operating in a silo. I focus on understanding the problem, making trade-offs explicit, and building solutions the team can trust.

Ownership without ego

I carry work from analysis and proof of concept through production rollout, testing, documentation, and handover. I ask for help when it is useful and share what I learn.

Calm under pressure

Tight deadlines and uncertainty call for a methodical approach. I stay focused on quality, risk, and delivery when several workstreams are moving in parallel.

A positive presence on the team

I contribute actively to technical discussions and bring concrete ideas. I enjoy working in teams where serious work and a relaxed atmosphere can coexist.

Case studies: critical production platforms

Production engineering

From proof of concept to operational handover: architecture, automation, production rollout, observability, and security for systems at scale.

Production-critical system

Perforce Backup System: replacing NetBackup across ~200 instances

2025 - 2026 · Architect and primary technical lead: proof of concept, architecture, development, production rollout, and handover

Distributed backup and recovery infrastructure: from a Kopia vs. Restic POC to production across ~200 Perforce instances (~1.3 PB source data, ~940 TB backed up).

  • ~200Perforce instances
  • ~1.3 PBsource data
  • ~940 TBbacked up

Technologies

  • Restic
  • Resticprofile
  • Python
  • Chef
  • systemd
  • Azure Blob Storage

View case study →

Production-critical system

Perforce Backup Replication: ~940 TB across four Azure regions

2026 · Architecture and development, commissioned by Security Risk Management

Cross-region recovery infrastructure: independent copies of ~940 TB of Restic backups across two Azure region pairs, with monthly delta synchronization.

  • ~940 TBof backups replicated
  • 4Azure regions
  • 41automated tests

Technologies

  • Python 3.13
  • Azure Functions
  • Durable Functions
  • AzCopy
  • Terraform
  • Azure Blob Storage

View case study →

Production-critical system

db-backup: Redis & MySQL backup platform

2021 - 2025 · Contributed to an existing project, supported production rollout, then assumed primary ownership of the service

Enterprise platform automating MySQL and Redis backup, storage, retention, validation, and restoration on Kubernetes.

  • ~190Redis/MySQL servers
  • 2database engines
  • ~US$200K/yrsavings (service-level)

Technologies

  • Python
  • Django REST
  • Go
  • Ansible
  • Kubernetes
  • Helm

View case study →

All case studies

Expertise

Cloud & Infrastructure

  • Linux
  • Azure (production)
  • Kubernetes (CKA)
  • Docker
  • OpenStack
  • Terraform
  • Helm

SRE & Observability

  • Prometheus
  • Grafana
  • Alertmanager
  • Azure Monitor / Log Analytics
  • Elasticsearch / Kibana
  • Incident response

Automation & Engineering

  • Python
  • Go
  • Bash
  • Ansible
  • Chef
  • GitLab CI/CD
  • REST APIs

Security

  • OIDC
  • Workload Identity Federation
  • Managed Identity
  • HashiCorp Vault
  • Secret management
  • Least privilege
  • Network isolation

Distributed Systems & Storage

  • Restic
  • Azure Blob Storage
  • Object storage
  • Petabyte-scale backup / recovery
  • Cross-region replication
  • Storage lifecycle / tiering
  • Fan-out / sharding

Personal Tooling & Automation

  • Claude Code
  • Coding agents

Open Source Contributions

Kopia

Open pull request adding Microsoft Entra ID (formerly Azure AD) authentication to Kopia's Azure Blob Storage backend through Managed Identity, Service Principal, and Azure CLI, with several iterations following code review.

Status : Open pull request

https://github.com/kopia/kopia/pull/4797

Earlier Experience

Education

Certifications & competitions

Professional Training

Languages

Contact