Running AI in Production: A Meetup for ML & AI Platform Engineers

Once you’re running AI in production, the hard problems are systems problems. GPUs idling at 5% utilization, and pinning down whether it’s data loading, comms overhead, or the scheduler is its own week of work. Distributed training runs that crash mid-job and leave you debugging whether it was the network, a flaky node, or NCCL acting up. Inference that benchmarks beautifully on a single request and collapses under concurrent load — KV cache pressure, batch dynamics, tail latency you can’t reproduce in staging. Stacks stitched together from vLLM, Ray, Triton, and Kubernetes that take more engineering hours to maintain than the model took to train.

None of this is the work you actually want to be doing. Every hour spent debugging NCCL or babysitting a Kubernetes cluster is an hour not spent on the model, the product, or the customer problem you’re actually paid to solve. Infrastructure should be the part that fades into the background, reliable enough that you trust it, fast enough that you don’t notice it, and out of the way so your team can ship.

Nebius and Anyscale are spending an evening in Stockholm on the patterns that actually address them.

On the agenda:

• From prototype to production: Reliable clusters for large scale distributed training

• Serverless AI on Nebius: run Jobs and Endpoints on demand for training, batch, and inference, with no cluster setup and no GPUs sitting idle between runs

• A technical customer deep-dive with Nebius Token Factory

• A presentation by Anyscale team

• Drinks, bites, and time to compare notes with people solving the same problems

Who should come

ML engineers, AI/platform infra teams, applied researchers, and technical founders working with GPUs, whether you’re scaling distributed training, fine-tuning open models, optimizing inference throughput, or just trying to keep production stable as load grows.

This is an evening with two of the teams whose tools are in your stack, in the city building some of Europe’s most interesting AI companies. Bring hard questions.