gramia

README.md



Cloud & AI platform engineer in Strasbourg. I build and run the infrastructure behind generative AI for an industrial company: containers, API gateways in front of LLMs, infrastructure as code and monitoring. I am also an AI business analyst and work with business teams to scope AI projects.
On my own time, I run a self-hosted LLM platform and write Grace Your Brain, a blog on Cloud, AI, DevOps and FinOps.
Currently

Building an agentic stack: K3s, a remote NVIDIA GPU, vLLM and an MCP server I use from VS Code.
Working through Under the hood, an 8-week study of Linux, containers, Kubernetes, networking, IaC, LLMs and observability. Notes and labs are public.
Writing the next articles for Grace Your Brain.

Selected work


Project notes
llm-platform: K3s cluster with a remote NVIDIA GPU reached over Tailscale. Models are served with vLLM (OpenAI-compatible API) and Open WebUI. GPU metrics go through DCGM Exporter, Prometheus and Grafana. A Python MCP server exposes Kubernetes, GPU and vLLM operations to VS Code.
cloud-landing-zone: Azure landing zone in Terraform and 2 AKS clusters (pre-production and production). I compared 5 to 8 tools for each pillar (FinOps, deployment, security, monitoring). OpenCost, Prometheus, Grafana and Argo CD were adopted.
datacenter: school project, team of 6. Routing and switching, DHCP/DDNS, OPNsense firewall, load balancing and reverse proxy, Ansible and Vagrant, Zabbix and Grafana, disaster recovery plan.
siem: log collection, centralized analysis and alerting for a private cloud, with OpenSearch and ElastAlert.

Toolbox

_{Also: Bicep, Container Apps, API Management, AI Foundry, Microsoft Fabric, Power BI, OpenCost, Open WebUI, OAuth 2.0, RBAC}
Certifications
   
_{Click a badge to verify it.}
Beyond the code

Outside work

Contact