Compose-to-Cloud Pulumi Providers for AWS, GCP, and Azure
Hi HN, These are three FOSS Pulumi providers (defang-aws, defang-gcp, defang-azure) that take a Docker Compose-like project and provision it as real infrastructure to the respective cloud: VPC, subnets, load balancers, serverless containers (ECS Fargate, Cloud Run, Container App), builds, managed PostgreSQL, Redis, LLMs, and public/private DNS. Develop once, deploy anywhere: all three providers expose the same Compose-shaped surface. They contain the same Pulumi resources: Project, Build, Service, Postgres, Redis, LLM. Each service in the Compose services map gets provisioned as one of…
What it does
In the maker’s words, at launch
Hi HN, These are three FOSS Pulumi providers (defang-aws, defang-gcp, defang-azure) that take a Docker Compose-like project and provision it as real infrastructure to the respective cloud: VPC, subnets, load balancers, serverless containers (ECS Fargate, Cloud Run, Container App), builds, managed PostgreSQL, Redis, LLMs, and public/private DNS. Develop once, deploy anywhere: all three providers expose the same Compose-shaped surface. They contain the same Pulumi resources: Project, Build, Service, Postgres, Redis, LLM. Each service in the Compose services map gets provisioned as one of those resources. Moving between clouds should be as simple as changing the import and nothing else. The Project resource sets up private DNS for service discovery, so Compose services can connect using their Compose service names. These providers have been rewritten from scratch in Go and are based on the (internal) TypeScript Pulumi code we've been working on for the last 2 years. We haven't reached parity quite yet, but we encourage everybody to use them and share your feedback. The Pulumi tooling generates SDKs for TypeScript, Python, Go, and C#. The repo has examples for all clouds in all those languages, eg.: const app = new defang_aws.Project("demo", { services: { web: { image: "nginx:latest", ports: [{ target: 80, mode: "ingress", appProtocol: "http" }], }, }, }); export const endpoints = app.endpoints; A few notes: - The Compose surfaces is opinionated. We check the service requirements and map it to a concrete cloud resource. For example, on GCP a service without exposed ports would run on Compose Engine, not Cloud Run. - Portability is at the Compose level. Each cloud's behaviour differs: IAM, native containers, etc. The API is the same, but the materialization isn't. - The Project resource owns the shared infra (VPC, build pipeline, DNS, LB). The standalone Service, Postgres, Redis resources skip that, so the Service needs a built image. - Licensing is dual: the provider engine is AGPL-3.0, because we want forks to contribute back to the community. The generated SDKs are Apache-2.0, so no copyleft in your own Pulumi code. - We're waiting for Pulumi devs to merge our registry PR, so the registry still reflects the old v1 provider docs. - There's support for runtime "config", ie. parameters, using a ConfigProvider interface, which also support interpolation. Repo: https://github.com/DefangLabs/pulumi-defang Happy to dig into how the Compose translation works, where the API diverges between clouds (and why), why Pulumi over CDK/Terraform, or the cloud-specific tradeoffs (AWS uses ECS+ALB, GCP uses Cloud Run with optional CE fallback, Azure uses Container Apps). - Lio
Does the same job
all alternatives →
Cloud World ModelJun 2026 · cloudworldmodel.ai · ▲172Simulate AWS, GCP & DigitalOcean without paying the bill
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 27d ago · cactuscompute.com
Launched alongside, May 2026
the whole month →
Parallel agents, diff reviewer, and multi-model comparisons
Dev tools · May 2026 · kilo.ai
- NW
Hey HN, Henry here from Cactus. We open-sourced Needle, a 26M parameter function-calling (tool use) model. It runs at 6000 tok/s prefill and 1200 tok/s decode on consumer devices. We were always frustrated by the little effort made towards building agentic models that run on budget phones, so we conducted investigations that led to an observation: agentic experiences are built upon tool calling, and massive models are overkill for it. Tool calling is fundamentally retrieval-and-assembly (match query to tool name, extract argument values, emit JSON), not reasoning. Cross-attention…
Life & fun · May 2026 · github.com
- FM
Dev tools · May 2026 · github.com