diff --git a/wsinfer_zoo/__init__.py b/wsinfer_zoo/__init__.py
index 650e4fd..ca73d95 100644
--- a/wsinfer_zoo/__init__.py
+++ b/wsinfer_zoo/__init__.py
@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 from wsinfer_zoo import _version
 from wsinfer_zoo import client  # noqa
 
diff --git a/wsinfer_zoo/__main__.py b/wsinfer_zoo/__main__.py
index 9acd8a1..7e56bb8 100644
--- a/wsinfer_zoo/__main__.py
+++ b/wsinfer_zoo/__main__.py
@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 from wsinfer_zoo.cli import cli
 
 if __name__ == "__main__":
diff --git a/wsinfer_zoo/cli.py b/wsinfer_zoo/cli.py
index 65a5eda..cb962bf 100644
--- a/wsinfer_zoo/cli.py
+++ b/wsinfer_zoo/cli.py
@@ -1,5 +1,7 @@
 """Command-line interface for the WSInfer model zoo."""
 
+from __future__ import annotations
+
 import dataclasses
 import json
 import sys
diff --git a/wsinfer_zoo/client.py b/wsinfer_zoo/client.py
index dab4df9..c7a030e 100644
--- a/wsinfer_zoo/client.py
+++ b/wsinfer_zoo/client.py
@@ -1,16 +1,14 @@
 """API to interact with WSInfer model zoo, hosted on HuggingFace."""
 
+from __future__ import annotations
+
 import dataclasses
 import functools
 import json
 import sys
 from pathlib import Path
 from typing import Any
-from typing import Dict
-from typing import List
-from typing import Optional
 from typing import Sequence
-from typing import Union
 
 import jsonschema
 from huggingface_hub import hf_hub_download
@@ -28,6 +26,7 @@
 WSINFER_ZOO_REGISTRY_DEFAULT_PATH = (
     Path.home() / ".wsinfer-zoo" / "wsinfer-zoo-registry.json"
 )
+WSINFER_ZOO_REGISTRY_DEFAULT_PATH.parent.mkdir(exist_ok=True)
 
 # In pyinstaller runtime for one-file executables, the root path
 # is the path to the executable.
@@ -93,7 +92,7 @@ class TransformConfigurationItem:
     """Container for one item in the 'transform' property of the model configuration."""
 
     name: str
-    arguments: Optional[Dict[str, Any]]
+    arguments: dict[str, Any] | None
 
 
 @dataclasses.dataclass
@@ -109,21 +108,21 @@ class ModelConfiguration:
     class_names: Sequence[str]
     patch_size_pixels: int
     spacing_um_px: float
-    transform: List[TransformConfigurationItem]
+    transform: list[TransformConfigurationItem]
 
     def __post_init__(self):
         if len(self.class_names) != self.num_classes:
             raise InvalidModelConfiguration()
 
     @classmethod
-    def from_dict(cls, config: Dict) -> "ModelConfiguration":
+    def from_dict(cls, config: dict) -> ModelConfiguration:
         validate_config_json(config)
         architecture = config["architecture"]
         num_classes = config["num_classes"]
         patch_size_pixels = config["patch_size_pixels"]
         spacing_um_px = config["spacing_um_px"]
         class_names = config["class_names"]
-        transform_list: List[Dict[str, Any]] = config["transform"]
+        transform_list: list[dict[str, Any]] = config["transform"]
         transform = [
             TransformConfigurationItem(name=t["name"], arguments=t.get("arguments"))
             for t in transform_list
@@ -157,7 +156,7 @@ class HFInfo:
     """Container for information on model's location on HuggingFace Hub."""
 
     repo_id: str
-    revision: Optional[str] = None
+    revision: str | None = None
 
 
 @dataclasses.dataclass
@@ -187,7 +186,7 @@ class HFModelWeightsOnly(HFModel):
 
 
 def load_torchscript_model_from_hf(
-    repo_id: str, revision: Optional[str] = None
+    repo_id: str, revision: str | None = None
 ) -> HFModelTorchScript:
     """Load a TorchScript model from HuggingFace."""
     model_path = hf_hub_download(repo_id, HF_TORCHSCRIPT_NAME, revision=revision)
@@ -207,7 +206,7 @@ def load_torchscript_model_from_hf(
 
 
 def load_weights_from_hf(
-    repo_id: str, revision: Optional[str] = None, safetensors: bool = False
+    repo_id: str, revision: str | None = None, safetensors: bool = False
 ) -> HFModelWeightsOnly:
     """Load model weights from HuggingFace (this is not TorchScript)."""
     if safetensors:
@@ -260,7 +259,7 @@ def __str__(self) -> str:
 class ModelRegistry:
     """Registry of models that can be used with WSInfer."""
 
-    models: Dict[str, RegisteredModel]
+    models: dict[str, RegisteredModel]
 
     def get_model_by_name(self, name: str) -> RegisteredModel:
         try:
@@ -269,7 +268,7 @@ def get_model_by_name(self, name: str) -> RegisteredModel:
             raise KeyError(f"model not found with name '{name}'.")
 
     @classmethod
-    def from_dict(cls, config: Dict) -> "ModelRegistry":
+    def from_dict(cls, config: dict) -> ModelRegistry:
         """Create a new ModelRegistry instance from a dictionary."""
         validate_model_zoo_json(config)
         models = {
@@ -286,7 +285,7 @@ def from_dict(cls, config: Dict) -> "ModelRegistry":
 
 
 @functools.lru_cache()
-def load_registry(registry_file: Optional[Union[str, Path]] = None) -> ModelRegistry:
+def load_registry(registry_file: str | Path | None = None) -> ModelRegistry:
     """Load model registry.
 
     This downloads the registry JSON file to a cache and reuses it if