feat: disable similarity worker for NER-only benchmark

main.py

@@ -321,40 +321,11 @@ async def analyze_resume(
     user_skills = [ent.text for ent in doc.ents if ent.label_ == "SKILL"]
     user_tools = [ent.text for ent in doc.ents if ent.label_ == "TOOL"]
 
-    # --- Similarity Search (via Isolated Subprocess) ---
-    # Fetch a large number of jobs to cache for pagination
+    # --- Similarity Search (DISABLED for NER-only benchmarking) ---
     all_similar_jobs = []
     total_similar_jobs = 0
-    try:
-        backend_dir = pathlib.Path(__file__).parent.resolve()
-        worker_path = backend_dir / "similarity_worker.py"
-        worker_python_executable = "/app/similarity_env/bin/python3"
-        
-        process = subprocess.run(
-            [
-                str(worker_python_executable),
-                str(worker_path),
-                "--target_role",
-                target_role or "Overall Market",
-                "--limit",
-                "200"  # Fetch a large batch for caching
-            ],
-            input=similarity_cleaned_text,
-            capture_output=True,
-            text=True,
-            check=True,
-        )
-        
-        worker_output = json.loads(process.stdout)
-        all_similar_jobs = worker_output.get("similar_jobs", [])
-        total_similar_jobs = worker_output.get("total_jobs", 0)
-
-    except (subprocess.CalledProcessError, json.JSONDecodeError, FileNotFoundError) as e:
-        print(f"Similarity worker failed: {e}", file=sys.stderr)
-        if isinstance(e, subprocess.CalledProcessError):
-            print(f"Worker stderr: {e.stderr}", file=sys.stderr)
-        all_similar_jobs = []
-        total_similar_jobs = 0
+    # The similarity worker subprocess call is bypassed for this benchmark.
+    # The original code for similarity search was here.
 
     # --- Cache the full results ---
     session_id = str(uuid.uuid4())