diff --git a/.env b/.env
new file mode 100644
index 0000000..7a7c0f9
--- /dev/null
+++ b/.env
@@ -0,0 +1 @@
+_APP_EMBEDDING_MODELS=embeddinggemma
\ No newline at end of file
diff --git a/docker-compose.yml b/docker-compose.yml
new file mode 100644
index 0000000..3ee539c
--- /dev/null
+++ b/docker-compose.yml
@@ -0,0 +1,19 @@
+---
+version: '3.8'
+
+services:
+  ollama:
+    build:
+      context: .
+      args:
+        MODELS: ${_APP_EMBEDDING_MODELS:-embeddinggemma}
+        # duration to keep model in memory
+        OLLAMA_KEEP_ALIVE: 24h
+    restart: unless-stopped
+    ports:
+      - "11434:11434"
+    volumes:
+      - appwrite-models:/root/.ollama
+    
+volumes:
+  appwrite-models:
\ No newline at end of file
diff --git a/dockerfile b/dockerfile
new file mode 100644
index 0000000..186a014
--- /dev/null
+++ b/dockerfile
@@ -0,0 +1,23 @@
+FROM ollama/ollama:0.12.7
+
+# Preload specific models
+ARG MODELS
+# needed to set in the environment
+ARG OLLAMA_KEEP_ALIVE
+ENV OLLAMA_KEEP_ALIVE=${OLLAMA_KEEP_ALIVE:-24h}
+
+# Pre-pull models at build time for Docker layer caching
+RUN ollama serve & \
+    sleep 5 && \
+    for m in $MODELS; do \
+        echo "Pulling model $m..."; \
+        ollama pull $m || exit 1; \
+    done && \
+    pkill ollama
+
+# Expose Ollama default port
+EXPOSE 11434
+
+# On container start, quickly ensure models exist (no re-download unless missing)
+ENTRYPOINT ["/bin/bash", "-c", "(sleep 2; for m in $MODELS; do ollama list | grep -q $m || ollama pull $m; done) & exec ollama $0"]
+CMD ["serve"]
\ No newline at end of file