Typical length
3–6 months
Four engagement shapes, scoped by what you actually need.
Every engagement is scope- and duration-dependent, so pricing is a conversation. These are the four shapes we actually work in.
01
EMBEDDED · 3–6 MONTHS
Platform Engineering Residency
A senior platform engineer embedded into your team to build — or rescue — an internal developer platform, Kubernetes fleet, or multi-cloud control surface.
What you get
This is the main thing we do. Residencies work when an organization has a real platform problem: developers fighting tooling, an IDP that’s stalled, a Kubernetes fleet that’s grown teeth, a multi-cloud migration halfway done. You get senior-level work — CAPI, Cilium, ArgoCD/Flux, Backstage, Crossplane, policy-as-code — delivered heads-down, with ADRs and runbooks as artifacts so the team stays unstuck after we leave.
We own delivery of what we're scoped to build: the architecture decisions, the implementation, and whether it actually ships. Your team owns the system afterward, which is why the handoff — runbooks, ADRs, an on-call playbook, pairing along the way — matters as much as the build itself. It's embedded, senior-level work with a clear owner, not a body added to your existing backlog.
Prior art · employment-eraLed at Alaska Airlines: an internal developer platform on Backstage with custom Go services, over a multi-cloud Kubernetes fleet spanning Azure, GCP and on-premises data centers.
02
BUILD · 2–6 MONTHS
AI Systems, Built Right
Agent architectures, evaluation harnesses, RAG that actually works, OCI-distributed skill patterns, built as working systems.
What you get
We build AI systems the way we build distributed systems — with observability, failure modes, eval harnesses, and a story for how the thing degrades. The pipelines we maintain internally (harness-go, agentisan, the OCI skill distribution work) are what we ship day to day, and we run local models alongside frontier ones on our own 128GB Strix Halo rig.
Engagements typically produce: one working system, an evaluation harness, a production runbook, and a clear answer to “what’s the local-model story?”
R&D exhibit · we own thisharness-go — autonomous multi-agent coding pipeline with NATS JetStream, BadgerDB + Bleve memory, hot-reload policy via atomic pointer swap during compaction.
03
ADVISORY · MONTHLY OR WORKSHOP
AI Advisory & Training
For leaders figuring out where AI actually fits, and engineering teams who want to stop pretending. Strategy, workshops, local-model evaluations.
What you get
A monthly retainer for a CTO or VP Eng who needs a senior technical ally thinking through AI strategy with them. Or a focused workshop: two hands-on days on agent architectures, a local-model evaluation with your real data, a regulated-AI deployment review.
The value is often in what we tell you to skip — the architecture that looks impressive and won’t survive an audit, the framework that solves a problem you don’t have yet.
What this looks likeA two-day on-site for a platform team: agent architectures, evaluation methodology, and a working harness prototype standing up by the end of day two — run against the same eval suite we use internally.
04
FIXED SCOPE · 2–6 WEEKS
Infra Rescue
The “your stack is on fire” short engagement. We show up, stabilize the thing, write down what to do differently, and leave.
What you get
Fixed-scope, time-boxed stabilization. An on-call runbook from hell, a Kubernetes upgrade that went sideways, a CI pipeline that takes 90 minutes and flakes, a region that keeps paging at 3am. We’ve seen it. We’ll stop the bleeding, document what we changed, and tell you what needs to change structurally so it doesn’t happen again.
Not a forever engagement. Two to six weeks, deliverables defined up front, and a post-mortem you can actually share with leadership.
What this looks likeFlakey GitOps sync across a multi-cluster fleet — resource contention, controller cardinality, and a missing leader-election config. Diagnosis, fix, and the write-up explaining why it happened.
Let's talk
Tell us what you're trying to ship —
we'll tell you what it looks like to get there.
We take one or two new engagements per quarter. No published pricing — every engagement is scope- and duration-dependent, so pricing is a conversation. Bring the problem. We'll bring the questions.