feat: kvbm + connector (#2258)

Signed-off-by: Ryan Olson <rolson@nvidia.com>
Co-authored-by: Olga Andreeva <oandreeva@nvidia.com>
Co-authored-by: Ziqi Fan <ziqif@nvidia.com>
Co-authored-by: John Thompson <jothomson@nvidia.com>
Co-authored-by: Richard Huo <rihuo@nvidia.com>
Co-authored-by: Zicheng Ma <zichengm@nvidia.com>

lib/bindings/python/src/dynamo/_core.pyi

@@ -1130,6 +1130,23 @@ class BlockManager:
         """
         ...
 
+class KvbmCacheManager:
+    """
+    A KV cache manager for VLLM
+    """
+
+    def __init__(self, block_manager: BlockManager) -> None:
+        ...
+
+
+class KvbmRequest:
+    """
+    A request for KV cache
+    """
+
+    def __init__(self, request_id: int, tokens: List[int], block_size: int) -> None:
+        ...
+
 class ZmqKvEventListener:
     """
     A ZMQ-based key-value cache event listener that operates independently

lib/bindings/python/src/dynamo/llm/__init__.py

@@ -9,6 +9,8 @@ from dynamo._core import AggregatedMetrics as AggregatedMetrics
 
 try:
     from dynamo._core import BlockManager as BlockManager
+    from dynamo._core import KvbmLeader as KvbmLeader
+    from dynamo._core import KvbmWorker as KvbmWorker
 except ImportError:
     pass  # BlockManager is not enabled by default