[ DEPLOY ] →
Your infra, production-ready in days.
We pick the hardware and deploy it via Terraform. Repo, IaC, secrets, monitoring and runbooks land as one package, not "docs later".
> embedded ops · devops · sre · dataops
Your DevOps team on contract. For Web3, AI, ZK and DePIN.
We know where to source hardware fast and how to ship nodes fast. We run your infra, so you don't hire SREs for four months only to lose them in eight.
team@zatva:~$
devops status --client=$YOU
@drxim
UTC+3
@shell
UTC+0
@serg
UTC-5
what we do
Three lifecycle phases. Each one works as a standalone engagement or as part of a retainer, whichever fits.
[ OPERATE ] →
On-call, patching, alerts: handled.
We own the pager, so you don't: PagerDuty rotations, a signed SLA, and a post-mortem after every Sev-1.
[ SCALE ] →
Grow 10× without downtime.
Cloud → bare-metal, region splits, hard-fork cutovers, burst capacity. We move running systems, and we do it reversibly.
who it's for
Planning a testnet launch next quarter? Still can't find an SRE who actually knows Cosmos SDK after months of looking, and the one you found wants equity on top? We get your validators live in 3 regions in 5 days, with slashing alerts and a signed uptime SLA.
Closed your round, bought the GPUs, and now the inference bill is bleeding faster than the product ships? Your ML team can train but doesn't want to get up at 3 AM for vLLM, OOMs and Triton? We run the inference layer (autoscaling, tracing, cost-per-token), so your researchers stop being on-call.
Proof generation is GPU-bound, deadline-bound and parallel all at once, so one dropped job and you miss the block. We build the farm, the queue, the retries, and per-circuit benchmarks on SP1 / RISC Zero / Boundless / Brevis.
The network pays for uptime, not excuses. Running 500 nodes across 10 regions by hand has turned into a part-time job no one on your team signed up for? We onboard, monitor and rebalance the fleet, and reconcile rewards weekly. We work with Filecoin, Akash, io.net, Render, Gensyn.
how we engage
01
Discovery
1 call. Scope, stack, regions, deadlines, SLA targets.
02
Plan
Deployment plan in 48h. Architecture, hardware selection, milestones, budget.
03
Deploy
Delivery via Terraform. Repo + IaC + monitoring + runbooks as one package.
04
Operate
Signed SLA. 24/7 on-call. Post-mortem after every Sev-1.
what we run for ourselves
We run our own fleet of 50+ nodes across 5+ countries, at 99.982% uptime. It's our proving ground: we break in the load, dashboards, runbooks and on-call rotations on infra we pay for ourselves, before any of it reaches you.
cases
slashing: 0 · downtime: 11 min/90d
cost/token: −60% · p95 latency: 380ms
uptime: 99.94% · reward tier: top-10%
top-5 operator · onboarding in 72h
stack we operate
Web3:
Cosmos SDK
Geth
Reth
OP Stack
Arbitrum Orbit
Polygon CDK
EigenDA
Celestia
AI / LLM:
vLLM
Triton
TensorRT-LLM
NVIDIA H100 / A100
Ray
Kubeflow
ZK:
SP1
RISC Zero
Boundless
Brevis
Jolt
Halo2
DePIN:
Filecoin
Akash
Render
io.net
Gensyn
Platform:
Kubernetes
Terraform
Ansible
Prometheus
Grafana
Loki
OpenTelemetry
PagerDuty
FAQ
No, we're a managed DevOps team. We deploy and operate infrastructure on the cloud or bare metal you own, and if you don't have your own hardware, we'll help you pick it and point you to where to rent. So you pay for the team, not the nodes.
A single L1/L2 validator runs 5-10 business days end-to-end. A GPU inference cluster across 3 regions takes 10-14 days. And a burst of up to 100 nodes for an incentivized testnet we turn around in 72h.
A mix: tier-1 clouds (AWS / GCP / Azure / Hetzner / OVH / Latitude), bare-metal partners (Latitude.sh, OpenMetal), and regional providers in 12+ countries. For each project we pick by latency, price and supply window.
Retainer plus project work in cash, and tokens are optional on a case-by-case basis. Equity-only engagements are the one thing we don't do.
You do. We work through an HSM/KMS workflow where keys never leave your control: we sign, we don't custody.
It's tiered. By default: p95 first response 15 min for Sev-1 and 1h for Sev-2, and higher tiers come with dedicated on-call.
Kubernetes-first, Terraform for IaC, Prometheus + Grafana + Loki for observability, PagerDuty for on-call. That said, we adapt to the client's environment.
The deployment plan has a fixed turnaround (48h), and the price depends on scope: send a short brief and we respond within 24h. Beyond that, it's hourly or monthly.
Often yes — it depends on region and GPU type. Send the spec and we'll quote the supply window within 24h.
NDA by default, and in public cases we anonymize the details.