Qualcomm AI Engine Direct - [GenAI Pipeline] PR2: Strategy Interfaces & Stage Wrappers (#20795)

## Summary

This PR adds the strategy interfaces, and stage wrappers for the GenAI
pipeline. This is the second of 7 PRs establishing the framework
skeleton. It builds on PR1 (data model + engine routing) and introduces
the Strategy pattern ABCs, concrete stage delegation wrappers and
Executorch strategy stubs. No existing files are modified (except
`__init__.py` introduced in PR1 for imports).

### What's included

#### Strategy ABCs (one class per file):

- `ModelPreparationStrategy` — in `strategies/model_preparation/`
- `QuantizationStrategy` — in `strategies/quantization/`
- `CompilationStrategy` — in `strategies/compilation/`
- `InferenceStrategy` — in `strategies/inference/`

Each ABC defines `invoke(context, input_config) -> output_config`.

#### Strategy stubs (raise `NotImplementedError`):

- `ExecuTorchModelPreparationStrategy`
- `ExecuTorchQuantizationStrategy`
- `ExecuTorchCompilationStrategy`
- `ExecuTorchInferenceStrategy`

#### Stage wrappers:

- `ModelPreparationStage`, `QuantizationStage`, `CompilationStage`,
`InferenceStage`.
- Each delegates to its injected strategy via `invoke()`.
- `name` property returns stage constant from `pipeline_types.py`.

#### Pipeline stage ABC:

- `PipelineStage` — abstract base with `name` property and `invoke()`
method.

#### Unit tests (16 tests across 6 test files):

- Strategy ABC enforcement (cannot instantiate, must implement
`invoke()`).
- Strategy stub tests (`NotImplementedError` raised).
- Stage delegation tests (`assert_called_once_with`).
- Stage name tests (use constants).

### PR Review Checklist

- All new classes follow single responsibility (one class per file) -
Yes.
- All dependencies are injected via constructor with sensible defaults -
Yes.
- All external calls are behind injectable interfaces - N/A, no external
calls.
- Unit tests cover every public method - Yes. 
- No existing files are modified (Phase 1 constraint) - Yes (only
`__init__.py` updated to add `PipelineStage` export).
- Docstrings on all public classes and methods - Yes. 
- Type annotations on all function signatures - Yes. 
- Logging follows the strategy in the LLD - N/A, no logging in this PR.

### Related PRs

- PR 1: Core data model, engine routing & exceptions:
#20409
- PR 2: Strategy interfaces & stage wrappers: this PR.
- PR 3: Pipeline orchestrator: pending.
- PR 4: Adapter interfaces & default implementations: pending.
- PR 5: Model preparation & quantization strategies: pending
- PR 6: Compilation & inference strategy implementations: pending.
- PR 7: Integration & E2E tests: pending.

## Test plan

```
python -m pytest \
  backends/qualcomm/genai_pipeline/tests/strategies/ \
  backends/qualcomm/genai_pipeline/tests/stages/ \
  -v
```

All 16 unit-tests passed.

### Test Coverage

Command to run:

```
coverage run --rcfile=backends/qualcomm/.coveragerc \
  -m pytest \
    backends/qualcomm/genai_pipeline/tests/strategies/ \
    backends/qualcomm/genai_pipeline/tests/stages/ \
  -v

coverage report --rcfile=backends/qualcomm/.coveragerc \
  --include="backends/qualcomm/genai_pipeline/pipeline_stage.py,backends/qualcomm/genai_pipeline/strategies/*,backends/qualcomm/genai_pipeline/stages/*"
```

Result:

```
Name                                                                                           Stmts   Miss Branch BrPart  Cover   Missing
------------------------------------------------------------------------------------------------------------------------------------------
backends/qualcomm/genai_pipeline/stages/compilation_stage.py                                      14      0      0      0   100%
backends/qualcomm/genai_pipeline/stages/inference_stage.py                                        14      0      0      0   100%
backends/qualcomm/genai_pipeline/stages/model_preparation_stage.py                                14      3      0      0    79%   24, 28, 35
backends/qualcomm/genai_pipeline/stages/quantization_stage.py                                     14      0      0      0   100%
backends/qualcomm/genai_pipeline/strategies/compilation/compilation_strategy.py                    7      0      0      0   100%
backends/qualcomm/genai_pipeline/strategies/compilation/executorch_compilation_strategy.py         7      0      0      0   100%
backends/qualcomm/genai_pipeline/strategies/inference/executorch_inference_strategy.py             7      0      0      0   100%
backends/qualcomm/genai_pipeline/strategies/inference/inference_strategy.py                        7      0      0      0   100%
backends/qualcomm/genai_pipeline/strategies/model_preparation/model_preparation_strategy.py        7      0      0      0   100%
backends/qualcomm/genai_pipeline/strategies/quantization/executorch_quantization_strategy.py       7      0      0      0   100%
backends/qualcomm/genai_pipeline/strategies/quantization/quantization_strategy.py                  7      0      0      0   100%
------------------------------------------------------------------------------------------------------------------------------------------
TOTAL                                                                                            110     3      0      0    97%
```

Author: qti-horodnic

backends/qualcomm/genai_pipeline/__init__.py

@@ -43,6 +43,7 @@
     "PipelineContext",
     "PipelineContextBuilder",
     "PipelineError",
+    "PipelineStage",
     "QuantizationInputConfig",
     "QuantizationOutputConfig",
     "StageError",

backends/qualcomm/genai_pipeline/pipeline_stage.py

@@ -0,0 +1,33 @@
+# Copyright (c) Qualcomm Innovation Center, Inc.
+# All rights reserved
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+from abc import ABC, abstractmethod
+from typing import Any
+
+
+class PipelineStage(ABC):
+    """Abstract base class for all pipeline stages.
+
+    Each stage delegates its work to an engine-specific strategy
+    resolved by the EngineProxy.
+    """
+
+    @property
+    @abstractmethod
+    def name(self) -> str: ...
+
+    @abstractmethod
+    def invoke(self, context: Any, input_config: Any) -> Any:
+        """Execute this pipeline stage.
+
+        Args:
+            context: The PipelineContext providing global pipeline settings.
+            input_config: The stage-specific input configuration dataclass.
+
+        Returns:
+            The stage-specific output configuration dataclass.
+        """
+        ...

backends/qualcomm/genai_pipeline/stages/__init__.py

@@ -0,0 +1,25 @@
+# Copyright (c) Qualcomm Innovation Center, Inc.
+# All rights reserved
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+from executorch.backends.qualcomm.genai_pipeline.stages.compilation_stage import (
+    CompilationStage,
+)
+from executorch.backends.qualcomm.genai_pipeline.stages.inference_stage import (
+    InferenceStage,
+)
+from executorch.backends.qualcomm.genai_pipeline.stages.model_preparation_stage import (
+    ModelPreparationStage,
+)
+from executorch.backends.qualcomm.genai_pipeline.stages.quantization_stage import (
+    QuantizationStage,
+)
+
+__all__ = [
+    "CompilationStage",
+    "InferenceStage",
+    "ModelPreparationStage",
+    "QuantizationStage",
+]

backends/qualcomm/genai_pipeline/stages/compilation_stage.py

@@ -0,0 +1,35 @@
+# Copyright (c) Qualcomm Innovation Center, Inc.
+# All rights reserved
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+from executorch.backends.qualcomm.genai_pipeline.configs.compilation_input_config import (
+    CompilationInputConfig,
+)
+from executorch.backends.qualcomm.genai_pipeline.configs.compilation_output_config import (
+    CompilationOutputConfig,
+)
+from executorch.backends.qualcomm.genai_pipeline.pipeline_context import PipelineContext
+from executorch.backends.qualcomm.genai_pipeline.pipeline_stage import PipelineStage
+from executorch.backends.qualcomm.genai_pipeline.pipeline_types import STAGE_COMPILATION
+from executorch.backends.qualcomm.genai_pipeline.strategies.compilation.compilation_strategy import (
+    CompilationStrategy,
+)
+
+
+class CompilationStage(PipelineStage):
+
+    def __init__(self, strategy: CompilationStrategy) -> None:
+        self._strategy = strategy
+
+    @property
+    def name(self) -> str:
+        return STAGE_COMPILATION
+
+    def invoke(
+        self,
+        context: PipelineContext,
+        input_config: CompilationInputConfig,
+    ) -> CompilationOutputConfig:
+        return self._strategy.invoke(context, input_config)

backends/qualcomm/genai_pipeline/stages/inference_stage.py

@@ -0,0 +1,35 @@
+# Copyright (c) Qualcomm Innovation Center, Inc.
+# All rights reserved
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+from executorch.backends.qualcomm.genai_pipeline.configs.inference_input_config import (
+    InferenceInputConfig,
+)
+from executorch.backends.qualcomm.genai_pipeline.configs.inference_output_config import (
+    InferenceOutputConfig,
+)
+from executorch.backends.qualcomm.genai_pipeline.pipeline_context import PipelineContext
+from executorch.backends.qualcomm.genai_pipeline.pipeline_stage import PipelineStage
+from executorch.backends.qualcomm.genai_pipeline.pipeline_types import STAGE_INFERENCE
+from executorch.backends.qualcomm.genai_pipeline.strategies.inference.inference_strategy import (
+    InferenceStrategy,
+)
+
+
+class InferenceStage(PipelineStage):
+
+    def __init__(self, strategy: InferenceStrategy) -> None:
+        self._strategy = strategy
+
+    @property
+    def name(self) -> str:
+        return STAGE_INFERENCE
+
+    def invoke(
+        self,
+        context: PipelineContext,
+        input_config: InferenceInputConfig,
+    ) -> InferenceOutputConfig:
+        return self._strategy.invoke(context, input_config)

backends/qualcomm/genai_pipeline/stages/model_preparation_stage.py

@@ -0,0 +1,37 @@
+# Copyright (c) Qualcomm Innovation Center, Inc.
+# All rights reserved
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+from executorch.backends.qualcomm.genai_pipeline.configs.model_preparation_input_config import (
+    ModelPreparationInputConfig,
+)
+from executorch.backends.qualcomm.genai_pipeline.configs.model_preparation_output_config import (
+    ModelPreparationOutputConfig,
+)
+from executorch.backends.qualcomm.genai_pipeline.pipeline_context import PipelineContext
+from executorch.backends.qualcomm.genai_pipeline.pipeline_stage import PipelineStage
+from executorch.backends.qualcomm.genai_pipeline.pipeline_types import (
+    STAGE_MODEL_PREPARATION,
+)
+from executorch.backends.qualcomm.genai_pipeline.strategies.model_preparation.model_preparation_strategy import (
+    ModelPreparationStrategy,
+)
+
+
+class ModelPreparationStage(PipelineStage):
+
+    def __init__(self, strategy: ModelPreparationStrategy) -> None:
+        self._strategy = strategy
+
+    @property
+    def name(self) -> str:
+        return STAGE_MODEL_PREPARATION
+
+    def invoke(
+        self,
+        context: PipelineContext,
+        input_config: ModelPreparationInputConfig,
+    ) -> ModelPreparationOutputConfig:
+        return self._strategy.invoke(context, input_config)

backends/qualcomm/genai_pipeline/stages/quantization_stage.py

@@ -0,0 +1,37 @@
+# Copyright (c) Qualcomm Innovation Center, Inc.
+# All rights reserved
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+from executorch.backends.qualcomm.genai_pipeline.configs.quantization_input_config import (
+    QuantizationInputConfig,
+)
+from executorch.backends.qualcomm.genai_pipeline.configs.quantization_output_config import (
+    QuantizationOutputConfig,
+)
+from executorch.backends.qualcomm.genai_pipeline.pipeline_context import PipelineContext
+from executorch.backends.qualcomm.genai_pipeline.pipeline_stage import PipelineStage
+from executorch.backends.qualcomm.genai_pipeline.pipeline_types import (
+    STAGE_QUANTIZATION,
+)
+from executorch.backends.qualcomm.genai_pipeline.strategies.quantization.quantization_strategy import (
+    QuantizationStrategy,
+)
+
+
+class QuantizationStage(PipelineStage):
+
+    def __init__(self, strategy: QuantizationStrategy) -> None:
+        self._strategy = strategy
+
+    @property
+    def name(self) -> str:
+        return STAGE_QUANTIZATION
+
+    def invoke(
+        self,
+        context: PipelineContext,
+        input_config: QuantizationInputConfig,
+    ) -> QuantizationOutputConfig:
+        return self._strategy.invoke(context, input_config)

backends/qualcomm/genai_pipeline/strategies/__init__.py

@@ -0,0 +1,25 @@
+# Copyright (c) Qualcomm Innovation Center, Inc.
+# All rights reserved
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+from executorch.backends.qualcomm.genai_pipeline.strategies.compilation.compilation_strategy import (
+    CompilationStrategy,
+)
+from executorch.backends.qualcomm.genai_pipeline.strategies.inference.inference_strategy import (
+    InferenceStrategy,
+)
+from executorch.backends.qualcomm.genai_pipeline.strategies.model_preparation.model_preparation_strategy import (
+    ModelPreparationStrategy,
+)
+from executorch.backends.qualcomm.genai_pipeline.strategies.quantization.quantization_strategy import (
+    QuantizationStrategy,
+)
+
+__all__ = [
+    "CompilationStrategy",
+    "InferenceStrategy",
+    "ModelPreparationStrategy",
+    "QuantizationStrategy",
+]

backends/qualcomm/genai_pipeline/strategies/compilation/__init__.py

@@ -0,0 +1,14 @@
+# Copyright (c) Qualcomm Innovation Center, Inc.
+# All rights reserved
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+from executorch.backends.qualcomm.genai_pipeline.strategies.compilation.compilation_strategy import (
+    CompilationStrategy,
+)
+from executorch.backends.qualcomm.genai_pipeline.strategies.compilation.executorch_compilation_strategy import (
+    ExecuTorchCompilationStrategy,
+)
+
+__all__ = ["CompilationStrategy", "ExecuTorchCompilationStrategy"]

backends/qualcomm/genai_pipeline/strategies/compilation/compilation_strategy.py

@@ -0,0 +1,34 @@
+# Copyright (c) Qualcomm Innovation Center, Inc.
+# All rights reserved
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+from abc import ABC, abstractmethod
+
+from executorch.backends.qualcomm.genai_pipeline.configs.compilation_input_config import (
+    CompilationInputConfig,
+)
+from executorch.backends.qualcomm.genai_pipeline.configs.compilation_output_config import (
+    CompilationOutputConfig,
+)
+from executorch.backends.qualcomm.genai_pipeline.pipeline_context import PipelineContext
+
+
+class CompilationStrategy(ABC):
+    @abstractmethod
+    def invoke(
+        self,
+        context: PipelineContext,
+        input_config: CompilationInputConfig,
+    ) -> CompilationOutputConfig:
+        """Compile the model to on-device artifacts.
+
+        Args:
+            context: The pipeline context with global settings.
+            input_config: The compilation input configuration.
+
+        Returns:
+            CompilationOutputConfig with artifact paths.
+        """
+        ...