Civo

🚨 We're hiring! Interested in joining the Civo team? We're looking for a Senior Solution Engineer – GPU & AI Infrastructure to lead the technical side of our large-scale AI and HPC deals, designing NVIDIA GPU clusters, InfiniBand/RoCE fabrics, and Kubernetes-based deployments for enterprise customers. This is a UK-based position, with a 4-day work week, uncapped holiday, remote work environment, and a genuinely collaborative team behind you. Apply now: https://civo.io/4qltsRI

Self-hosting isn't just about cutting subscription costs anymore, it's about who controls your data. Abdul Talha put together a full walkthrough for deploying AFFiNE, an open-source docs and whiteboard workspace, on Civo Kubernetes. The setup includes a pgvector-enabled managed PostgreSQL database and Redis caching, so semantic search and AI copilot features run without a single byte leaving your own infrastructure. For teams weighing sovereignty against convenience, this is what having both actually looks like in practice. Read the deployment guide: https://civo.io/4hRkeuh

Every time a developer pauses mid-function, their editor quietly ships the surrounding code to an external server. For teams working on proprietary algorithms or regulated data, that's not a convenience, it's a liability. Mostafa Ibrahim wrote a hands-on tutorial to deploying Tabby, a self-hosted code completion server, on Civo L40S GPUs. Qwen2.5-Coder 7B runs entirely inside your own Kubernetes cluster: ghost-text completions in VS Code, ready in under an hour, with your code never leaving infrastructure you control. Read the full tutorial: https://civo.io/4wGBOWr

Platform team trying to enforce standards without becoming the bottleneck? Engineering lead tired of cluster config being something only one person on the team actually understands? 👀 On 19 August, John Dietz and M R Rishi show exactly what a well-structured template strategy looks like with Konstruct: built-in templates across Civo, AWS, GCP, Azure, and bare metal, plus your own custom templates backed by your Terraform modules and GitOps content. Build the strategy you'll still trust six months from now, not the one that quietly falls apart the moment someone leaves the team. Save your seat: https://civo.io/4yX8ywp

Blackwell or Blackwell Ultra: does the upgrade actually change your decision? The NVIDIA B300 and B200 share an architecture but not a performance profile. This comparison lays out the differences in memory, compute, and where each one makes the most sense, so you're not guessing which GPU your next workload needs: https://civo.io/4vPKKrj

🚨 We're hiring! Interested in joining the Civo team? We're looking for a Senior Solution Engineer – GPU & AI Infrastructure to lead the technical side of our large-scale AI and HPC deals, designing NVIDIA GPU clusters, InfiniBand/RoCE fabrics, and Kubernetes-based deployments for enterprise customers. This is a UK-based position, with a 4-day work week, uncapped holiday, remote work environment, and a genuinely collaborative team behind you. Apply now: https://civo.io/4xJBwhT

Benchmarks are easy to publish. Surviving a GPU eviction mid-inference is not. This deep dive benchmarks vLLM inference on Civo H100 GPUs, covering throughput, what actually happens during eviction recovery, and a full cost analysis so you know what production really costs, not just what the demo looks like. Read the tutorial from Mostafa Ibrahim: https://civo.io/4pKkZXY

Public cloud freedom or private cloud control. Why should you have to pick one? CivoStack Enterprise and FlexCore both give you a private cloud with full public cloud parity, but they solve for different priorities. This breakdown compares the two so you can work out which one actually fits how your organization wants to run its infrastructure. Get more from this blog from Russ Smith: https://civo.io/4xvMVBV

Fine-tuning a frontier model doesn't require a data center of your own 🤔 Using LoRA and Unsloth, this tutorial from Mostafa Ibrahim shows how to fine-tune Llama 4 on a single Civo H100 cluster, cutting the compute and memory overhead while keeping the workflow reproducible enough for a real production pipeline. Read the full tutorial: https://civo.io/4vQZMwZ

It’s the classic cloud trap: you start with a hyperscaler because it's fast and easy. Fast forward a year, and your team's assumptions, services, and architecture are heavily dependent on proprietary tools. So, what triggers a switch? And how do the numbers actually look when you move away from the big hyperscalers? Catch our upcoming webinar on September 9th with KubeNine Founder Vikas Yadav and Civo’s Deepak Kadira. We’re stripping away the vendor pitch to take a peer-to-peer look at how cloud costs compound, when to start looking at alternatives, and why the "ops toil" of self-hosted/managed services is no longer as expensive as it used to be. Grab your seat and join the conversation: https://lnkd.in/eNkCVvey #CloudArchitecture #Kubernetes #FinOps #CloudNative #TechLeadership #Webinar #india