Cloud Infrastructure
Everything as code. AWS-first, cloud-agnostic by design. Networking, identity, compute and data planned in pull requests and applied by a pipeline, never from a laptop.
Terraform · multi-account · migrations
Platform engineering · Cloud · AI · Product
Infrastructure, snapped to perfection.
SnapOps is a hands-on platform engineering consultancy. Cloud infrastructure, the pipelines that ship to it, the developer platform on top, the AI tooling inside the workflow, and the products that run on all of it. The whole lot, from one engineer who does the work.
Everything lands in your accounts and your repositories. When we're done, you own it.
Services
Six disciplines. One pair of hands.
Each is a practice on its own. Together they are a platform. Fixed scope where the work is known, a retainer where it isn't; either way you get code in your repositories and a document your next engineer can follow.
AI Engineering & Tooling
LLM tools that survive production: agents, retrieval, structured outputs, evaluations and guardrails. AI inside the developer workflow, measured before it is trusted.
Agents · RAG · evals · guardrails
Internal Developer Platform
The paved road your engineers actually take: self-service environments, service templates, secrets and access done once and done right.
Golden paths · self-service · templates
CI/CD & Release
Build, test, deploy, promote and roll back with the fewest moving parts that still fail safely. GitOps where it helps, not where it impresses.
GitHub Actions · GitOps · rollbacks
Reliability & Cost
Knowing what happened, and what it cost, before a customer tells you. Paging for the right things, an on-call story people can live with, a bill explained line by line.
Observability · on-call · FinOps
Product Engineering
The thing on top of the platform. Web, mobile and backend, built end to end and shipped to real users. BuyAway is what that looks like when it is ours.
Web · mobile · backend
All six, in one hand.
Fractional platform leadership for teams that need a head of platform's judgement without the headcount: roadmap, build-versus-buy, vendor and cloud decisions, hiring plans, and the architecture reviews that keep all six pointing the same way.
AWSTerraformKubernetesECS & LambdaGitHub ActionsDockerCloudflarePostgresSupabaseLLM APIsPythonTypeScriptFlutter
How I work
Review, build, hand over.
No account managers, no juniors learning on your bill. You talk to the person doing the work, from the first design call to the last commit.
01
Review
A short, fixed-price look at where you are. What is fragile, what is expensive, what is fine and should be left alone. Written down, in the order I would fix things.
02
Build
Agreed scope, agreed price, delivered as pull requests into your repositories. Every change explains itself, so the reasoning survives after I leave.
03
Hand over, or stay
Your team runs it, with the documentation to do so. Or a light retainer covers upgrades, reviews and the occasional 2 a.m.
One rule. Nothing lives only on my machine or in my accounts. Credentials in your secrets manager, code in your repos, state in your buckets. Lock-in is a smell.
About
A one-person consultancy, by design.
SnapOps is run by Haim Elbaz, a platform engineer based in Israel who leads platform engineering by day and builds the rest of the time, working with clients remotely wherever they are. It stays small on purpose: the person you brief is the person who writes the code, and the person who answers when it breaks.
Most engagements start with a review and end with a hand-over. The best outcome is a team that no longer needs me, a platform that is boring to operate, and one more thing in the world that just works when you snap your fingers.
Shipped under our own nameBuyAway · Is it cheaper here, or back home? A price-comparison app for travellers.buyaway.snapops.cloud →Contact
Ready to snap your platform into place?
Tell me what you are running, what you are building, and what keeps you up at night. A first conversation is free and usually enough to know whether I can help.