diff --git a/ai_stack/llamacpp.server/README.md b/ai_stack/llamacpp.server/README.md
index 303af9a..450c8cf 100644
--- a/ai_stack/llamacpp.server/README.md
+++ b/ai_stack/llamacpp.server/README.md
@@ -22,6 +22,9 @@ Native way to run as a persistent background service on Debian/Ubuntu.
   2. Reload systemd: `sudo systemctl daemon-reload`
   3. Enable and Start: `sudo systemctl enable --now llama-server`
 - **Check health**: `sudo systemctl status llama-server`
+- **View logs**: 
+  - Real-time: `journalctl -u llama-server -f`
+  - Since boot: `journalctl -u llama-server -b`
 
 ---
 
diff --git a/ai_stack/llamacpp.server/llama-server.service b/ai_stack/llamacpp.server/llama-server.service
index 817c403..8895a9e 100644
--- a/ai_stack/llamacpp.server/llama-server.service
+++ b/ai_stack/llamacpp.server/llama-server.service
@@ -4,8 +4,8 @@ After=network.target
 
 [Service]
 Type=simple
-Environment="MODEL_PATH=/home/user/unsloth/Qwen3.5-0.8B-GGUF/Qwen3.5-0.8B-UD-Q4_K_XL.gguf"
-Environment="MMPROJ_PATH=/home/user/unsloth/Qwen3.5-0.8B-GGUF/mmproj-BF16.gguf"
+Environment="MODEL_PATH=%h/unsloth/Qwen3.5-0.8B-GGUF/Qwen3.5-0.8B-UD-Q4_K_XL.gguf"
+Environment="MMPROJ_PATH=%h/unsloth/Qwen3.5-0.8B-GGUF/mmproj-BF16.gguf"
 ExecStart=/usr/local/bin/llama-server \
   --model ${MODEL_PATH} \
   --mmproj ${MMPROJ_PATH} \
@@ -21,10 +21,10 @@ ExecStart=/usr/local/bin/llama-server \
   --chat-template-kwargs '{"enable_thinking":false}'
 Restart=always
 RestartSec=5
-User=user
-WorkingDirectory=/home/user/projects/home_stack/ai_stack/llamacpp.server
-StandardOutput=append:/home/user/projects/home_stack/ai_stack/llamacpp.server/server.log
-StandardError=append:/home/user/projects/home_stack/ai_stack/llamacpp.server/server.log
+User=%u
+WorkingDirectory=%h/projects/home_stack/ai_stack/llamacpp.server
+StandardOutput=journal
+StandardError=journal
 
 [Install]
 WantedBy=multi-user.target