feat: AI 서비스 MLX 듀얼 백엔드 및 모델 최적화

- MLX(맥미니 27B) 우선 → Ollama(조립컴 9B) fallback 구조
- pydantic-settings 기반 config 전환
- health check에 MLX 상태 추가
- 텍스트 모델 qwen3:8b → qwen3.5:9b-q8_0 변경

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

ai-service/services/ollama_client.py

@@ -1,3 +1,4 @@
+import asyncio
 import httpx
 from config import settings
 
@@ -6,29 +7,55 @@ class OllamaClient:
     def __init__(self):
         self.base_url = settings.OLLAMA_BASE_URL
         self.timeout = httpx.Timeout(float(settings.OLLAMA_TIMEOUT), connect=10.0)
+        self._client: httpx.AsyncClient | None = None
+
+    async def _get_client(self) -> httpx.AsyncClient:
+        if self._client is None or self._client.is_closed:
+            self._client = httpx.AsyncClient(timeout=self.timeout)
+        return self._client
+
+    async def close(self):
+        if self._client and not self._client.is_closed:
+            await self._client.aclose()
+            self._client = None
 
     async def generate_embedding(self, text: str) -> list[float]:
-        async with httpx.AsyncClient(timeout=self.timeout) as client:
-            response = await client.post(
-                f"{self.base_url}/api/embeddings",
-                json={"model": settings.OLLAMA_EMBED_MODEL, "prompt": text},
-            )
-            response.raise_for_status()
-            return response.json()["embedding"]
+        client = await self._get_client()
+        response = await client.post(
+            f"{self.base_url}/api/embeddings",
+            json={"model": settings.OLLAMA_EMBED_MODEL, "prompt": text},
+        )
+        response.raise_for_status()
+        return response.json()["embedding"]
 
-    async def batch_embeddings(self, texts: list[str]) -> list[list[float]]:
-        results = []
-        for text in texts:
-            emb = await self.generate_embedding(text)
-            results.append(emb)
-        return results
+    async def batch_embeddings(self, texts: list[str], concurrency: int = 5) -> list[list[float]]:
+        semaphore = asyncio.Semaphore(concurrency)
+
+        async def _embed(text: str) -> list[float]:
+            async with semaphore:
+                return await self.generate_embedding(text)
+
+        return await asyncio.gather(*[_embed(t) for t in texts])
 
     async def generate_text(self, prompt: str, system: str = None) -> str:
         messages = []
         if system:
             messages.append({"role": "system", "content": system})
         messages.append({"role": "user", "content": prompt})
-        async with httpx.AsyncClient(timeout=self.timeout) as client:
+        client = await self._get_client()
+        try:
+            response = await client.post(
+                f"{settings.MLX_BASE_URL}/chat/completions",
+                json={
+                    "model": settings.MLX_TEXT_MODEL,
+                    "messages": messages,
+                    "max_tokens": 2048,
+                    "temperature": 0.3,
+                },
+            )
+            response.raise_for_status()
+            return response.json()["choices"][0]["message"]["content"]
+        except Exception:
             response = await client.post(
                 f"{self.base_url}/api/chat",
                 json={
@@ -42,16 +69,21 @@ class OllamaClient:
             return response.json()["message"]["content"]
 
     async def check_health(self) -> dict:
+        result = {}
         try:
-            async with httpx.AsyncClient(timeout=httpx.Timeout(5.0)) as client:
-                response = await client.get(f"{self.base_url}/api/tags")
-                models = response.json().get("models", [])
-                return {
-                    "status": "connected",
-                    "models": [m["name"] for m in models],
-                }
+            client = await self._get_client()
+            response = await client.get(f"{self.base_url}/api/tags")
+            models = response.json().get("models", [])
+            result["ollama"] = {"status": "connected", "models": [m["name"] for m in models]}
         except Exception:
-            return {"status": "disconnected"}
+            result["ollama"] = {"status": "disconnected"}
+        try:
+            client = await self._get_client()
+            response = await client.get(f"{settings.MLX_BASE_URL}/health")
+            result["mlx"] = {"status": "connected", "model": settings.MLX_TEXT_MODEL}
+        except Exception:
+            result["mlx"] = {"status": "disconnected"}
+        return result
 
 
 ollama_client = OllamaClient()