diff --git a/README.md b/README.md
index 1001791..834f122 100644
--- a/README.md
+++ b/README.md
@@ -29,6 +29,14 @@ $ deadline bundle gui-submit job_bundles/gui_control_showcase
 
 ![deadline bundle gui-submit showcase](.images/deadline-bundle-gui-submit-showcase.png)
 
+## Container samples
+
+The [containers](containers) directory contains Dockerfiles for building
+container images compatible with Deadline Cloud worker environments. The
+[al2023-deadline](containers/al2023-deadline/) sample replicates the
+service-managed fleet worker AMI package set on Amazon Linux 2023, useful for
+building and testing conda packages or other software locally.
+
 ## Conda recipes
 
 The [conda_recipes](conda_recipes) directory contains samples and tooling for building conda packages for your
diff --git a/containers/README.md b/containers/README.md
new file mode 100644
index 0000000..f8bdc8d
--- /dev/null
+++ b/containers/README.md
@@ -0,0 +1,18 @@
+# AWS Deadline Cloud container samples
+
+The container samples in this directory provide Dockerfiles and related resources
+for building container images compatible with
+[AWS Deadline Cloud](https://aws.amazon.com/deadline-cloud/) worker environments.
+
+Use these to build and test software locally with the same system libraries,
+toolchains, and runtime environment as Deadline Cloud workers.
+
+## Samples
+
+### AL2023 worker-equivalent image
+
+The [al2023-deadline](al2023-deadline/) sample provides a Dockerfile that
+replicates the package set of the Deadline Cloud service-managed fleet (SMF)
+worker AMI on top of the base Amazon Linux 2023 image. Use it to build and test
+[conda packages](../conda_recipes/) or other software that must be compatible
+with the worker runtime.
diff --git a/containers/al2023-deadline/README.md b/containers/al2023-deadline/README.md
new file mode 100644
index 0000000..9b698c6
--- /dev/null
+++ b/containers/al2023-deadline/README.md
@@ -0,0 +1,70 @@
+# AL2023 Deadline Cloud worker-equivalent image
+
+This Dockerfile replicates the package set of an April 2026 snapshot of the
+AWS Deadline Cloud service-managed fleet (SMF) worker AMI on top of the base
+Amazon Linux 2023 image. It is a point-in-time capture — the actual worker AMI
+may drift as packages are added or updated.
+
+## Use cases
+
+- Build and test [conda packages](../../conda_recipes/) with the same GLIBC
+  version, system libraries, and runtime environment as real workers.
+- Reproduce worker-side build or runtime failures locally.
+- Validate that your software dependencies are satisfied by the worker
+  environment before submitting jobs.
+
+## What's included
+
+The image installs packages in layered groups matching the worker AMI:
+
+| Layer | Contents |
+|-------|----------|
+| Core system tools | `jq`, `git`, `wget`, `unzip`, `vim`, `sudo`, `rsync`, … |
+| Build toolchain | GCC, G++, `binutils`, `kernel-headers`, `glibc-devel`, `zlib-devel` |
+| X11 / Mesa / OpenGL | Headless rendering dependencies (`libX11`, `mesa-libGL`, `libglvnd`, …) |
+| Image / media libs | `libjpeg-turbo`, `libpng`, `libtiff`, `libwebp` |
+| Networking / NFS / security | NFS utils, NSS, SSSD, `openssh-clients`, `iptables-nft` |
+| Python 3.11 | `python3.11`, `pip`, `setuptools` |
+| Docker / containerd | For container-in-container workflows (host socket mount or `--privileged`) |
+| Misc | AWS CLI v2, Boost, jemalloc, TBB, and other libraries present on the worker |
+
+## Building the image
+
+```bash
+docker build -f Dockerfile.worker-equivalent -t al2023-deadline:latest .
+```
+
+## Running a container
+
+```bash
+# Interactive shell
+docker run --rm -it al2023-deadline:latest
+
+# Build a conda package inside the container
+docker run --rm -v "$PWD":/work -w /work al2023-deadline:latest \
+    bash -c "pip3.11 install conda-build && conda build my-recipe/"
+```
+
+## GPU support
+
+For NVIDIA GPU support, add the NVIDIA container toolkit repository before
+building:
+
+```dockerfile
+RUN dnf config-manager --add-repo \
+    https://nvidia.github.io/libnvidia-container/stable/rpm/nvidia-container-toolkit.repo \
+ && dnf install -y nvidia-container-toolkit
+```
+
+Then run the container with `--gpus all`:
+
+```bash
+docker run --rm --gpus all al2023-deadline:latest nvidia-smi
+```
+
+## Limitations
+
+- This is a **point-in-time snapshot** (April 2026). The actual SMF worker AMI
+  may have newer or additional packages.
+- Docker-in-Docker is not enabled by default. Mount the host Docker socket or
+  use `--privileged` if you need it.