Services
From infrastructure to production, without surprises
Three areas, the ones of my experience. They all start the same way: understanding what exists before changing anything.
Kubernetes platform
Kubernetes clusters built, secured and run
I install and maintain your clusters, from test to production: upgrades, scaling, deployments without downtime. Everything is described in code, so that your team can take it over.
For whom: Teams that host their applications on their own servers or with a cloud provider (OVHcloud, AWS, GCP, Azure).
What you get
- Clusters installed with Kubespray, one per environment (DEV, TEST, PREPROD, PROD)
- Rolling and canary deployments, HPA and VPA scaling
- GitOps with Argo CD: Git describes what runs
- Security: RBAC, NetworkPolicy, Pod Security "restricted", automated TLS certificates
Done already: At Hoortrade, the DEV, TEST, PREPROD and PROD clusters of the SaaS platform; on the Kubernetes Lab project, a two-node cluster run and documented. See the project →
CI/CD and DevSecOps
Pipelines that ship, and block what must be blocked
I design your GitLab CI/CD pipelines end to end: build, tests, security scan, deployment. An unscanned image does not reach production, and every change goes through a review.
For whom: Development teams that want to ship more often without breaking production.
What you get
- GitLab CI/CD pipelines: build, tests, deployment
- Image scanning (Trivy) before promotion
- GitLab administration: access, branching strategies, code review
- Infrastructure as Code with Terraform and Ansible
Done already: At Hoortrade, the pipelines and the GitLab administration; on this site, a pipeline that builds, scans, promotes, deploys and checks. See the project →
Observability and SRE
See outages before your users do
I set up the monitoring and the alerts that matter, then I keep the platform running: incidents, root cause analyses, backups proven by restores.
For whom: Teams that learn about outages from their customers, or drown in useless alerts.
What you get
- Prometheus, Alertmanager and Grafana, with readable dashboards
- Useful alerts, by e-mail or Slack, with their runbook
- Incident management and root cause analyses (RCA)
- Site availability, and backups proven by restores
Done already: At Hoortrade, Prometheus, Grafana and Zabbix across the platform, incidents and RCA; on my projects, the availability of every site and centralised logs. See the project →
How it works
A first exchange
You describe the need through the form. I answer myself, and tell you honestly whether I am the right person.
An inventory
I look at what exists before proposing anything: what runs, what breaks, what costs.
Reviewed changes
Every change goes through a merge request and a green pipeline. Nothing reaches production by hand.
An autonomous team
Everything is described in code and documented, limits included. Your team can take over.
In practice
See my projects →- Contract, engagement or project?
- All three: a one-off contract, a longer engagement or a whole project. Describe the need and I will tell you how I can help.
- Where?
- Based in Montréal. On site in Montréal, remote elsewhere.
- How can I check?
- What I describe runs online: the projects and the documentation are public.
- What if Kubernetes is not the right answer?
- I will tell you. A well-run server or a managed service is often enough, and costs much less.
A DevOps / SRE project or contract?
Describe the need in a few lines. I answer myself, by e-mail.
Send a message