feat(frontend): First part of Python request handling (#4999)

Signed-off-by: Graham King <grahamk@nvidia.com>

components/src/dynamo/frontend/main.py

@@ -30,12 +30,15 @@ from dynamo.llm import (
     EngineType,
     EntrypointArgs,
     KvRouterConfig,
+    ModelDeploymentCard,
+    PythonAsyncEngine,
     RouterConfig,
     RouterMode,
     make_engine,
     run_input,
 )
 from dynamo.runtime import DistributedRuntime
+from dynamo.runtime.logging import configure_dynamo_logging
 
 from . import __version__
 
@@ -45,9 +48,25 @@ CUSTOM_BACKEND_METRICS_POLLING_INTERVAL_ENV_VAR = (
 )
 CUSTOM_BACKEND_ENDPOINT_ENV_VAR = "CUSTOM_BACKEND_ENDPOINT"
 
+configure_dynamo_logging()
 logger = logging.getLogger(__name__)
 
 
+async def _dummy_generator(request):
+    """Minimal generator that yields nothing. Work in progress."""
+    return
+    yield  # Makes this an async generator
+
+
+async def engine_factory(mdc: ModelDeploymentCard) -> PythonAsyncEngine:
+    """
+    Called by Rust when a model is discovered.
+    """
+    loop = asyncio.get_running_loop()
+    logger.info(f"Engine_factory called with MDC: {mdc.to_json_str()[:100]}...")
+    return PythonAsyncEngine(_dummy_generator, loop)
+
+
 def validate_model_name(value):
     """Validate that model-name is a non-empty string."""
     if not value or not isinstance(value, str) or len(value.strip()) == 0:
@@ -254,6 +273,12 @@ def parse_args():
         default=os.environ.get("DYN_REQUEST_PLANE", "nats"),
         help="Determines how requests are distributed from routers to workers. 'tcp' is fastest [nats|http|tcp]",
     )
+    parser.add_argument(
+        "--exp-python-factory",
+        action="store_true",
+        default=False,
+        help="[EXPERIMENTAL] Enable Python-based engine factory. When set, engines will be created via a Python callback instead of the default Rust pipeline.",
+    )
 
     flags = parser.parse_args()
 
@@ -356,6 +381,9 @@ async def async_main():
             "custom_backend_metrics_polling_interval"
         ] = flags.custom_backend_metrics_polling_interval
 
+    if flags.exp_python_factory:
+        kwargs["engine_factory"] = engine_factory
+
     e = EntrypointArgs(EngineType.Dynamic, **kwargs)
     engine = await make_engine(runtime, e)
 

launch/dynamo-run/src/lib.rs

@@ -146,7 +146,10 @@ async fn engine_for(
     match out_opt {
         Output::Auto => {
             // Auto-discover backends
-            Ok(EngineConfig::Dynamic(Box::new(local_model)))
+            Ok(EngineConfig::Dynamic {
+                model: Box::new(local_model),
+                engine_factory: None,
+            })
         }
         Output::Echo => Ok(EngineConfig::InProcessText {
             model: Box::new(local_model),

lib/bindings/c/src/lib.rs

@@ -975,6 +975,7 @@ pub async fn create_worker_selection_pipeline_chat(
         component.drt().clone(),
         model_manager.clone(),
         router_config,
+        None,
     );
     let cards = watcher
         .cards_for_model(model_name, Some(namespace), false)

lib/bindings/python/rust/llm/entrypoint.rs

@@ -2,20 +2,29 @@
 // SPDX-License-Identifier: Apache-2.0
 
 use std::fmt::Display;
+use std::future::Future;
 use std::path::PathBuf;
+use std::pin::Pin;
+use std::sync::Arc;
 
 use pyo3::{exceptions::PyException, prelude::*};
+use pyo3_async_runtimes::TaskLocals;
 
 use dynamo_llm::entrypoint::EngineConfig as RsEngineConfig;
+use dynamo_llm::entrypoint::EngineFactoryCallback;
 use dynamo_llm::entrypoint::RouterConfig as RsRouterConfig;
 use dynamo_llm::entrypoint::input::Input;
 use dynamo_llm::kv_router::KvRouterConfig as RsKvRouterConfig;
 use dynamo_llm::local_model::DEFAULT_HTTP_PORT;
 use dynamo_llm::local_model::{LocalModel, LocalModelBuilder};
 use dynamo_llm::mocker::protocols::MockEngineArgs;
+use dynamo_llm::model_card::ModelDeploymentCard as RsModelDeploymentCard;
+use dynamo_llm::types::openai::chat_completions::OpenAIChatCompletionsStreamingEngine;
 use dynamo_runtime::protocols::EndpointId;
 
+use super::model_card::ModelDeploymentCard;
 use crate::RouterMode;
+use crate::engine::PythonAsyncEngine;
 
 #[pyclass(eq, eq_int)]
 #[derive(Clone, Debug, PartialEq)]
@@ -117,6 +126,21 @@ impl From<RouterConfig> for RsRouterConfig {
     }
 }
 
+/// Wrapper to hold Python callback and its TaskLocals for async execution
+#[derive(Clone)]
+struct PyEngineFactory {
+    callback: Arc<PyObject>,
+    locals: Arc<TaskLocals>,
+}
+
+impl std::fmt::Debug for PyEngineFactory {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("PyEngineFactory")
+            .field("callback", &"<PyObject>")
+            .finish()
+    }
+}
+
 #[pyclass]
 #[derive(Clone, Debug)]
 pub(crate) struct EntrypointArgs {
@@ -137,14 +161,16 @@ pub(crate) struct EntrypointArgs {
     custom_backend_metrics_endpoint: Option<String>,
     custom_backend_metrics_polling_interval: Option<f64>,
     is_prefill: bool,
+    engine_factory: Option<PyEngineFactory>,
 }
 
 #[pymethods]
 impl EntrypointArgs {
     #[allow(clippy::too_many_arguments)]
     #[new]
-    #[pyo3(signature = (engine_type, model_path=None, model_name=None, endpoint_id=None, context_length=None, template_file=None, router_config=None, kv_cache_block_size=None, http_host=None, http_port=None, tls_cert_path=None, tls_key_path=None, extra_engine_args=None, namespace=None, custom_backend_metrics_endpoint=None, custom_backend_metrics_polling_interval=None, is_prefill=false))]
+    #[pyo3(signature = (engine_type, model_path=None, model_name=None, endpoint_id=None, context_length=None, template_file=None, router_config=None, kv_cache_block_size=None, http_host=None, http_port=None, tls_cert_path=None, tls_key_path=None, extra_engine_args=None, namespace=None, custom_backend_metrics_endpoint=None, custom_backend_metrics_polling_interval=None, is_prefill=false, engine_factory=None))]
     pub fn new(
+        py: Python<'_>,
         engine_type: EngineType,
         model_path: Option<PathBuf>,
         model_name: Option<String>, // e.g. "dyn://namespace.component.endpoint"
@@ -162,6 +188,7 @@ impl EntrypointArgs {
         custom_backend_metrics_endpoint: Option<String>,
         custom_backend_metrics_polling_interval: Option<f64>,
         is_prefill: bool,
+        engine_factory: Option<PyObject>,
     ) -> PyResult<Self> {
         let endpoint_id_obj: Option<EndpointId> = endpoint_id.as_deref().map(EndpointId::from);
         if (tls_cert_path.is_some() && tls_key_path.is_none())
@@ -171,6 +198,23 @@ impl EntrypointArgs {
                 "tls_cert_path and tls_key_path must be provided together",
             ));
         }
+
+        // Capture TaskLocals at registration time for the engine factory callback
+        let engine_factory = engine_factory
+            .map(|callback| {
+                let locals = pyo3_async_runtimes::tokio::get_current_locals(py).map_err(|e| {
+                    pyo3::exceptions::PyRuntimeError::new_err(format!(
+                        "Failed to get TaskLocals for engine_factory: {}",
+                        e
+                    ))
+                })?;
+                Ok::<_, PyErr>(PyEngineFactory {
+                    callback: Arc::new(callback),
+                    locals: Arc::new(locals),
+                })
+            })
+            .transpose()?;
+
         Ok(EntrypointArgs {
             engine_type,
             model_path,
@@ -189,6 +233,7 @@ impl EntrypointArgs {
             custom_backend_metrics_endpoint,
             custom_backend_metrics_polling_interval,
             is_prefill,
+            engine_factory,
         })
     }
 }
@@ -251,6 +296,57 @@ pub fn make_engine<'p>(
     })
 }
 
+/// Convert a PyEngineFactory to a Rust EngineFactoryCallback
+fn py_engine_factory_to_callback(factory: PyEngineFactory) -> EngineFactoryCallback {
+    let callback = factory.callback;
+    let locals = factory.locals;
+
+    Arc::new(
+        move |card: RsModelDeploymentCard| -> Pin<
+            Box<dyn Future<Output = anyhow::Result<OpenAIChatCompletionsStreamingEngine>> + Send>,
+        > {
+            let callback = callback.clone();
+            let locals = locals.clone();
+
+            Box::pin(async move {
+                // Acquire GIL to call Python callback and convert coroutine to future
+                let py_future = Python::with_gil(|py| {
+                    // Create Python ModelDeploymentCard wrapper
+                    let py_card = ModelDeploymentCard { inner: card };
+                    let py_card_obj = Py::new(py, py_card)
+                        .map_err(|e| anyhow::anyhow!("Failed to create Python MDC: {}", e))?;
+
+                    // Call Python async function to get a coroutine
+                    let coroutine = callback
+                        .call1(py, (py_card_obj,))
+                        .map_err(|e| anyhow::anyhow!("Failed to call engine_factory: {}", e))?;
+
+                    // Use the TaskLocals captured at registration time
+                    pyo3_async_runtimes::into_future_with_locals(&locals, coroutine.into_bound(py))
+                        .map_err(|e| {
+                            anyhow::anyhow!("Failed to convert coroutine to future: {}", e)
+                        })
+                })?;
+
+                // Await the Python coroutine (GIL is released during await)
+                let py_result = py_future
+                    .await
+                    .map_err(|e| anyhow::anyhow!("engine_factory callback failed: {}", e))?;
+
+                // Extract PythonAsyncEngine from the Python result and wrap in Arc
+                let engine: OpenAIChatCompletionsStreamingEngine = Python::with_gil(|py| {
+                    let engine: PythonAsyncEngine = py_result.extract(py).map_err(|e| {
+                        anyhow::anyhow!("Failed to extract PythonAsyncEngine: {}", e)
+                    })?;
+                    Ok::<_, anyhow::Error>(Arc::new(engine))
+                })?;
+
+                Ok(engine)
+            })
+        },
+    )
+}
+
 async fn select_engine(
     #[allow(unused_variables)] distributed_runtime: super::DistributedRuntime,
     args: EntrypointArgs,
@@ -264,7 +360,14 @@ async fn select_engine(
                 engine: dynamo_llm::engines::make_echo_engine(),
             }
         }
-        EngineType::Dynamic => RsEngineConfig::Dynamic(Box::new(local_model)),
+        EngineType::Dynamic => {
+            //  Convert Python engine factory to Rust callback
+            let engine_factory = args.engine_factory.map(py_engine_factory_to_callback);
+            RsEngineConfig::Dynamic {
+                model: Box::new(local_model),
+                engine_factory,
+            }
+        }
         EngineType::Mocker => {
             let mocker_args = if let Some(extra_args_path) = args.extra_engine_args {
                 MockEngineArgs::from_json_file(&extra_args_path).map_err(|e| {

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

@@ -21,6 +21,7 @@ from dynamo._core import KvStats as KvStats
 from dynamo._core import LoRADownloader as LoRADownloader
 from dynamo._core import MediaDecoder as MediaDecoder
 from dynamo._core import MediaFetcher as MediaFetcher
+from dynamo._core import ModelDeploymentCard as ModelDeploymentCard
 from dynamo._core import ModelInput as ModelInput
 from dynamo._core import ModelRuntimeConfig as ModelRuntimeConfig
 from dynamo._core import ModelType as ModelType

lib/llm/src/discovery/watcher.rs

@@ -20,7 +20,7 @@ use dynamo_runtime::{
 
 use crate::{
     backend::Backend,
-    entrypoint::{self, RouterConfig},
+    entrypoint::{self, EngineFactoryCallback, RouterConfig},
     kv_router::PrefillRouter,
     model_card::ModelDeploymentCard,
     model_type::{ModelInput, ModelType},
@@ -53,6 +53,7 @@ pub struct ModelWatcher {
     router_config: RouterConfig,
     notify_on_model: Notify,
     model_update_tx: Option<Sender<ModelUpdate>>,
+    engine_factory: Option<EngineFactoryCallback>,
 }
 
 const ALL_MODEL_TYPES: &[ModelType] = &[
@@ -68,6 +69,7 @@ impl ModelWatcher {
         runtime: DistributedRuntime,
         model_manager: Arc<ModelManager>,
         router_config: RouterConfig,
+        engine_factory: Option<EngineFactoryCallback>,
     ) -> ModelWatcher {
         Self {
             manager: model_manager,
@@ -75,6 +77,7 @@ impl ModelWatcher {
             router_config,
             notify_on_model: Notify::new(),
             model_update_tx: None,
+            engine_factory,
         }
     }
 
@@ -355,6 +358,7 @@ impl ModelWatcher {
         if let Some(tx) = &self.model_update_tx {
             tx.send(ModelUpdate::Added(card.clone())).await.ok();
         }
+
         let checksum = card.mdcsum();
 
         if card.model_input == ModelInput::Tokens
@@ -429,7 +433,13 @@ impl ModelWatcher {
 
             // Add chat engine only if the model supports chat
             if card.model_type.supports_chat() {
-                let chat_engine = entrypoint::build_routed_pipeline::<
+                // Work in progress. This will allow creating  a chat_engine from Python.
+                let chat_engine = if let Some(ref factory) = self.engine_factory {
+                    factory(card.clone())
+                        .await
+                        .context("python engine_factory")?
+                } else {
+                    entrypoint::build_routed_pipeline::<
                         NvCreateChatCompletionRequest,
                         NvCreateChatCompletionStreamResponse,
                     >(
@@ -443,7 +453,8 @@ impl ModelWatcher {
                         self.router_config.enforce_disagg,
                     )
                     .await
-                .context("build_routed_pipeline")?;
+                    .context("build_routed_pipeline")?
+                };
                 self.manager
                     .add_chat_completions_model(card.name(), checksum, chat_engine)
                     .context("add_chat_completions_model")?;

lib/llm/src/entrypoint.rs

@@ -8,15 +8,28 @@
 pub mod input;
 pub use input::{build_routed_pipeline, build_routed_pipeline_with_preprocessor};
 
+use std::future::Future;
+use std::pin::Pin;
 use std::sync::Arc;
 
 use dynamo_runtime::pipeline::RouterMode;
 
 use crate::{
     backend::ExecutionContext, engines::StreamingEngine, kv_router::KvRouterConfig,
-    local_model::LocalModel,
+    local_model::LocalModel, model_card::ModelDeploymentCard,
+    types::openai::chat_completions::OpenAIChatCompletionsStreamingEngine,
 };
 
+/// Callback type for engine factory (async)
+pub type EngineFactoryCallback = Arc<
+    dyn Fn(
+            ModelDeploymentCard,
+        ) -> Pin<
+            Box<dyn Future<Output = anyhow::Result<OpenAIChatCompletionsStreamingEngine>> + Send>,
+        > + Send
+        + Sync,
+>;
+
 #[derive(Debug, Clone, Default)]
 pub struct RouterConfig {
     pub router_mode: RouterMode,
@@ -58,7 +71,10 @@ impl RouterConfig {
 #[derive(Clone)]
 pub enum EngineConfig {
     /// Remote networked engines that we discover via etcd
-    Dynamic(Box<LocalModel>),
+    Dynamic {
+        model: Box<LocalModel>,
+        engine_factory: Option<EngineFactoryCallback>,
+    },
 
     /// A Text engine receives text, does it's own tokenization and prompt formatting.
     InProcessText {
@@ -75,12 +91,19 @@ pub enum EngineConfig {
 }
 
 impl EngineConfig {
-    fn local_model(&self) -> &LocalModel {
+    pub fn local_model(&self) -> &LocalModel {
         use EngineConfig::*;
         match self {
-            Dynamic(lm) => lm,
+            Dynamic { model, .. } => model,
             InProcessText { model, .. } => model,
             InProcessTokens { model, .. } => model,
         }
     }
+
+    pub fn engine_factory(&self) -> Option<&EngineFactoryCallback> {
+        match self {
+            EngineConfig::Dynamic { engine_factory, .. } => engine_factory.as_ref(),
+            _ => None,
+        }
+    }
 }

lib/llm/src/entrypoint/input/common.rs

@@ -58,12 +58,15 @@ pub async fn prepare_engine(
     engine_config: EngineConfig,
 ) -> anyhow::Result<PreparedEngine> {
     match engine_config {
-        EngineConfig::Dynamic(local_model) => {
+        EngineConfig::Dynamic {
+            model: local_model, ..
+        } => {
             let model_manager = Arc::new(ModelManager::new());
             let watch_obj = Arc::new(ModelWatcher::new(
                 distributed_runtime.clone(),
                 model_manager.clone(),
                 RouterConfig::default(),
+                None,
             ));
             let discovery = distributed_runtime.discovery();
             let discovery_stream = discovery

lib/llm/src/entrypoint/input/endpoint.rs

@@ -82,7 +82,7 @@ pub async fn run(
             let fut = endpoint.endpoint_builder().handler(ingress).start();
             Box::pin(fut)
         }
-        EngineConfig::Dynamic(_) => {
+        EngineConfig::Dynamic { .. } => {
             unreachable!("An endpoint input will never have a Dynamic engine");
         }
     };

lib/llm/src/entrypoint/input/grpc.rs

@@ -26,11 +26,11 @@ pub async fn run(
         .with_request_template(engine_config.local_model().request_template());
 
     let grpc_service = match engine_config {
-        EngineConfig::Dynamic(_) => {
+        EngineConfig::Dynamic { ref model, .. } => {
             let grpc_service = grpc_service_builder.build()?;
-            let router_config = engine_config.local_model().router_config();
+            let router_config = model.router_config();
             // Listen for models registering themselves, add them to gRPC service
-            let namespace = engine_config.local_model().namespace().unwrap_or("");
+            let namespace = model.namespace().unwrap_or("");
             let target_namespace = if is_global_namespace(namespace) {
                 None
             } else {
@@ -105,7 +105,7 @@ async fn run_watcher(
     router_config: RouterConfig,
     target_namespace: Option<String>,
 ) -> anyhow::Result<()> {
-    let watch_obj = ModelWatcher::new(runtime.clone(), model_manager, router_config);
+    let watch_obj = ModelWatcher::new(runtime.clone(), model_manager, router_config, None);
     tracing::debug!("Waiting for remote model");
     let discovery = runtime.discovery();
     let discovery_stream = discovery

lib/llm/src/entrypoint/input/http.rs

@@ -7,7 +7,7 @@ use crate::{
     discovery::{ModelManager, ModelUpdate, ModelWatcher},
     endpoint_type::EndpointType,
     engines::StreamingEngineAdapter,
-    entrypoint::{EngineConfig, RouterConfig, input::common},
+    entrypoint::{EngineConfig, EngineFactoryCallback, RouterConfig, input::common},
     http::service::service_v2::{self, HttpService},
     namespace::is_global_namespace,
     types::openai::{
@@ -61,16 +61,19 @@ pub async fn run(
     );
 
     let http_service = match engine_config {
-        EngineConfig::Dynamic(_) => {
+        EngineConfig::Dynamic {
+            ref model,
+            ref engine_factory,
+        } => {
             // This allows the /health endpoint to query store for active instances
             http_service_builder = http_service_builder.store(distributed_runtime.store().clone());
             let http_service = http_service_builder.build()?;
 
-            let router_config = engine_config.local_model().router_config();
+            let router_config = model.router_config();
             // Listen for models registering themselves, add them to HTTP service
             // Check if we should filter by namespace (based on the local model's namespace)
             // Get namespace from the model, fallback to endpoint_id namespace if not set
-            let namespace = engine_config.local_model().namespace().unwrap_or("");
+            let namespace = model.namespace().unwrap_or("");
             let target_namespace = if is_global_namespace(namespace) {
                 None
             } else {
@@ -83,6 +86,7 @@ pub async fn run(
                 target_namespace,
                 Arc::new(http_service.clone()),
                 http_service.state().metrics_clone(),
+                engine_factory.clone(),
             )
             .await?;
             http_service
@@ -194,8 +198,14 @@ async fn run_watcher(
     target_namespace: Option<String>,
     http_service: Arc<HttpService>,
     metrics: Arc<crate::http::service::metrics::Metrics>,
+    engine_factory: Option<EngineFactoryCallback>,
 ) -> anyhow::Result<()> {
-    let mut watch_obj = ModelWatcher::new(runtime.clone(), model_manager, router_config);
+    let mut watch_obj = ModelWatcher::new(
+        runtime.clone(),
+        model_manager,
+        router_config,
+        engine_factory,
+    );
     tracing::debug!("Waiting for remote model");
     let discovery = runtime.discovery();
     let discovery_stream = discovery

lib/llm/tests/http_metrics.rs

@@ -339,6 +339,7 @@ mod integration_tests {
             distributed_runtime.clone(),
             service.state().manager_clone(),
             dynamo_llm::entrypoint::RouterConfig::default(),
+            None,
         );
         // Start watching for model registrations via discovery interface
         let discovery = distributed_runtime.discovery();
@@ -510,6 +511,7 @@ mod integration_tests {
                 distributed_runtime.clone(),
                 service.state().manager_clone(),
                 dynamo_llm::entrypoint::RouterConfig::default(),
+                None,
             );
 
             // Get all model entries for our test model

lib/runtime/src/system_status_server.rs

@@ -961,7 +961,7 @@ mod integration_tests {
 
                 // Now create a namespace, component, and endpoint to make the system healthy
                 let namespace = drt.namespace("ns1234").unwrap();
-                let mut component = namespace.component("comp1234").unwrap();
+                let component = namespace.component("comp1234").unwrap();
 
                 // Create a simple test handler
                 use crate::pipeline::{async_trait, network::Ingress, AsyncEngine, AsyncEngineContextProvider, Error, ManyOut, SingleIn};