First Platform Engineering Meetup NL
I am excited to announce that I am speaking at the inaugural Platform Engineering Meetup NL at Xebia in Amsterdam — Wibautstraat 200 — on Wednesday, April 22, 2026.
This is the first edition of a community meetup focused on the Dutch platform engineering scene, and the lineup brings together two topics that are increasingly relevant for modern engineering organizations: controlling infrastructure cost and operating shared AI platforms at scale.
My Talk: Lessons Learned Orchestrating Multi-Tenant GPUs on OpenShift AI
How do you run shared, production-grade GPUs for AI/ML safely and efficiently?
In this experience report, I share lessons learned from implementing multi-tenant GPU orchestration on OpenShift AI using NVIDIA KAI on G/H200 hardware, fronted by Traefik and backed by Dell Technologies platforms.
What I Cover
- Tenant isolation patterns — namespaces, quotas, and priority classes for safe multi-tenancy
- Scheduling across heterogeneous nodes — mixing GPU types without manual node selection
- MIG versus full-GPU trade-offs — when to partition, when to dedicate
- Throughput versus latency tuning — optimizing for inference SLAs vs training throughput
- Driver and firmware pitfalls — the bugs you only find in production
- Upgrade and rollback strategies — zero-downtime GPU driver updates
- Day-2 operations — observability, autoscaling, and chargeback models
This talk builds directly on my KubeCon Europe 2026 presentation on multi-tenant GPUs and my Red Hat Summit 2026 lightning talk — but goes deeper into the operational lessons that only emerge after months of running GPU workloads in production.
Also Speaking: Cloud Cost Optimization at Scale
The evening kicks off with Darko Klincharski (DevOps Engineering Lead at Tarmac.io) presenting “Data Hoarder’s Guide to Surviving Cloud Bills” — a real-world case study on reducing AWS spend for a platform serving 700 million users. Practical strategies for managing petabytes of static data, high-traffic volumes, and large database cleanup.
What the evening looked like
I wrote the sections above before the meetup. Here is what the slides in the room showed on the night. The decks carried the branding “Cloud Native Platform NL” and the footer “Cloud Native Platform Meetup, Amsterdam, 22 April 2026”. Darko’s talk was the first session of the evening.

Darko’s talk in progress, with the “Move Away from S3” slide on both screens.
The talk opened with a cartoon title slide, “Data Hoarder’s Guide To Surviving Cloud Bills” by Darko Klincharski, who goes by “DarkOps” on the slide.

The title slide, as projected.
One of the later slides was titled “Move Away from S3”: a table comparing object-storage offerings on price, with the source credited to s3compare.io. My take: before you tune anything else in a cloud bill, check what you pay per stored terabyte against the alternatives.

The “Move Away from S3” comparison table.
The slide that closed Darko’s session was headed “Thanks, Darko!” with a feedback form code and an “Up next” panel announcing my talk, “Multi-Tenant GPUs on OpenShift AI with NVIDIA KAI”, subtitled “Lessons learned from G/H200 hardware, Traefik-fronted, on Dell platforms”. After my talk a closing slide asked “How was the GPU talk?” and invited the room to scan a code with feedback for the next edition.
Event Details
| Detail | Info |
|---|---|
| Event | Platform Engineering Meetup NL — Edition 1 |
| Date | Wednesday, April 22, 2026 |
| Time | 18:00 - 21:00 |
| Location | Xebia, Wibautstraat 200, Amsterdam |
| Food | Pizza and drinks provided |
Agenda
- 18:00 — Arrival with pizza and drinks
- 18:45 — Talks and discussion
- 20:30 — Last round of drinks
- 21:00 — Building closes
Who Should Attend
- Platform engineers and SREs
- DevOps engineers and cloud architects
- Engineering managers and CTOs
- Anyone working on Kubernetes, cost optimization, developer platforms, or AI infrastructure
Why This Matters
Platform engineering is no longer just about CI/CD pipelines and developer portals. The rise of AI workloads has added a new dimension: GPU orchestration is now a platform engineering problem. Teams need to share expensive hardware across multiple tenants, enforce quotas, provide self-service access, and deliver observability — all while keeping costs under control.
This meetup sits at exactly that intersection: cloud cost management meets AI platform operations.
If you are in Amsterdam, come say hello. I would love to discuss GPU multi-tenancy, platform engineering patterns, or anything cloud native.
Want to discuss GPU platform engineering for your organization? I help enterprises design multi-tenant GPU platforms on Kubernetes and OpenShift.

