AI 및 fastAPI 초기 구조 셋팅

ai-server/backend/ai/embedding.py

@@ -0,0 +1,9 @@
+from langchain_ibm import WatsonxEmbeddings
+from backend.config.settings import settings
+
+watson_embedding = WatsonxEmbeddings(
+    model_id="ibm/granite-embedding-278m-multilingual",
+    url=f"{settings.watsonx_url}",
+    api_key=f"{settings.watsonx_api_key}",
+    project_id=f"{settings.watsonx_project_id}",
+)

ai-server/backend/ai/llm.py

@@ -0,0 +1,17 @@
+from langchain_ibm import ChatWatsonx
+from backend.config.settings import settings
+from langchain_openai import ChatOpenAI
+
+watson_llm = ChatWatsonx(
+    model_id="ibm/granite-4-h-small",
+    url=f"{settings.watsonx_url}",
+    api_key=f"{settings.watsonx_api_key}",
+    project_id=f"{settings.watsonx_project_id}",
+    max_tokens=2000,
+)
+
+hugging_llm = ChatOpenAI(
+    base_url="https://router.huggingface.co/v1",
+    api_key=f"{settings.hf_token}",
+    model_name="Qwen/Qwen3-8B:nscale",
+)