diff --git a/services/mam-api/src/routes/recorders.js b/services/mam-api/src/routes/recorders.js
index 04ca3ef..acfd698 100644
--- a/services/mam-api/src/routes/recorders.js
+++ b/services/mam-api/src/routes/recorders.js
@@ -789,43 +789,22 @@ router.post('/:id/stop', requireRecorderEdit, async (req, res, next) => {
         return res.status(502).json({ error: 'Remote node failed to stop sidecar' });
       }
     } else {
-      // Issue #162 — stop WITH a grace period (t=180). Docker sends SIGTERM and
-      // waits up to 180s for ffmpeg to flush and write the MOV/MP4 trailer before
-      // it SIGKILLs. Without this the master is truncated/corrupt and the
-      // pre-created asset can get stuck in 'live'.
-      const stopRes = await dockerApi(
-        'POST',
-        `/containers/${recorder.container_id}/stop?t=180`
-      );
-
-      // 204 = stopped, 304 = already stopped, 404 = container gone — all acceptable.
-      if (stopRes.status !== 204 && stopRes.status !== 304 && stopRes.status !== 404) {
-        return res.status(500).json({
-          error: 'Failed to stop container',
-          details: stopRes.data,
-        });
-      }
-
-      // Only attempt remove if the container existed (not 404).
-      if (stopRes.status !== 404) {
-        // Issue #162 — before removing the container, wait for the master to
-        // finalize (asset leaves 'live'), mirroring the remote path's reliance on
-        // the node-agent's clean teardown. This guards against deleting the
-        // container — and its lingering finalize work — too early.
-        await waitForFinalize(recorder);
-
-        const removeRes = await dockerApi(
-          'DELETE',
-          `/containers/${recorder.container_id}`
-        );
-
-        if (removeRes.status !== 204 && removeRes.status !== 404) {
-          return res.status(500).json({
-            error: 'Failed to remove container',
-            details: removeRes.data,
-          });
+      // Issue #162 — stop local container in the background so the HTTP stop
+      // request returns immediately. The container teardown (SIGTERM -> ffmpeg
+      // exit -> S3 upload -> post-stop callback) takes up to 180s for large files,
+      // which would otherwise timeout the browser/API connection.
+      const containerId = recorder.container_id;
+      (async () => {
+        try {
+          const stopRes = await dockerApi('POST', `/containers/${containerId}/stop?t=180`);
+          if (stopRes.status !== 404) {
+            await waitForFinalize(recorder);
+            await dockerApi('DELETE', `/containers/${containerId}?force=true`).catch(() => {});
+          }
+        } catch (e) {
+          console.error('[recorders] failed local background stop:', e.message);
         }
-      }
+      })();
     }
 
     const updateResult = await pool.query(
diff --git a/services/node-agent/index.js b/services/node-agent/index.js
index f702c3b..fc71219 100644
--- a/services/node-agent/index.js
+++ b/services/node-agent/index.js
@@ -387,29 +387,33 @@ async function fetchContainerLogs(containerId) {
 async function handleSidecarStop(containerId, res) {
   try {
     console.log(`[sidecar-stop] stopping ${containerId} (grace 180s)...`);
-    // Grace period must exceed the capture container's shutdown work
-    // (finalise ffmpeg session + register asset via callback). Default
-    // docker stop is only 10s, which SIGKILLs capture mid-finalise and
-    // loses the POST /assets callback -> asset stuck 'live', no jobs.
-    await dockerApi('POST', `/containers/${containerId}/stop?t=180`).catch(() => {});
-    // Dump the capture container's shutdown logs into our persistent log
-    // BEFORE removing it, so failed callbacks are diagnosable.
-    const logs = await fetchContainerLogs(containerId);
-    console.log(`[sidecar-stop] ==== capture logs for ${containerId} ====\n${logs}\n[sidecar-stop] ==== end logs ====`);
-    // Container has now exited gracefully (or hit the 180s cap); remove it.
-    await dockerApi('DELETE', `/containers/${containerId}?force=true`).catch(() => {});
+    
+    // Run the container teardown and cleanup in the background. The capture
+    // process SIGTERM handler flushes ffmpeg and uploads the file to S3
+    // (taking up to 3 minutes for multi-GB files) before exiting. Returning
+    // immediately stops the API request timing out.
+    (async () => {
+      try {
+        await dockerApi('POST', `/containers/${containerId}/stop?t=180`).catch(() => {});
+        const logs = await fetchContainerLogs(containerId);
+        console.log(`[sidecar-stop] ==== capture logs for ${containerId} ====\n${logs}\n[sidecar-stop] ==== end logs ====`);
+        await dockerApi('DELETE', `/containers/${containerId}?force=true`).catch(() => {});
 
-    // Deltacast bridge lifecycle: decrement sidecar count; stop bridge when last.
-    if (_containerSourceType.get(containerId) === 'deltacast') {
-      _containerSourceType.delete(containerId);
-      _dcSidecarCount--;
-      if (_dcSidecarCount <= 0) {
-        _dcSidecarCount = 0;
-        stopDeltacastBridge();
+        // Deltacast bridge lifecycle: decrement sidecar count; stop bridge when last.
+        if (_containerSourceType.get(containerId) === 'deltacast') {
+          _containerSourceType.delete(containerId);
+          _dcSidecarCount--;
+          if (_dcSidecarCount <= 0) {
+            _dcSidecarCount = 0;
+            stopDeltacastBridge();
+          }
+        } else {
+          _containerSourceType.delete(containerId);
+        }
+      } catch (err) {
+        console.error(`[sidecar-stop] background cleanup failed for ${containerId}:`, err.message);
       }
-    } else {
-      _containerSourceType.delete(containerId);
-    }
+    })();
 
     jsonResponse(res, 200, { ok: true });
   } catch (err) {