Process AI jobs concurrently

internal/worker/worker.go

@@ -5,6 +5,7 @@ import (
 	"encoding/json"
 	"log/slog"
 	"strings"
+	"sync"
 	"time"
 
 	"ai-service/internal/llm"
@@ -94,6 +95,18 @@ func (w *Worker) tick(ctx context.Context) {
 		slog.Error("claim jobs failed", "error", err)
 		return
 	}
+	if len(jobs) > 1 {
+		var wg sync.WaitGroup
+		wg.Add(len(jobs))
+		for _, job := range jobs {
+			go func(job *model.Job) {
+				defer wg.Done()
+				w.process(ctx, job)
+			}(job)
+		}
+		wg.Wait()
+		return
+	}
 	for _, job := range jobs {
 		w.process(ctx, job)
 	}

k8s/worker-deployment.yaml

@@ -31,6 +31,8 @@ spec:
               value: "llm_chat,chat_completion,call_analysis,telegram_classification"
             - name: WORKER_MODEL_PROFILES
               value: "qwen2.5-14b"
+            - name: WORKER_CLAIM_LIMIT
+              value: "8"
           envFrom:
             - configMapRef:
                 name: ai-service-config