Model Engine OnPrem Support and vLLM 0.11.1 + Model Engine Integration Fixes#744

Open

charlesahn-scale wants to merge 30 commits intomainfrom

onPremSuport

Commits on Feb 3, 2026

add support for on-prem

TarunRavikumar
authored and
charlesahn-scale
committed
clean up on-prem artificats

TarunRavikumar
authored and
charlesahn-scale
committed
add back comments from initial code

TarunRavikumar
authored and
charlesahn-scale
committed
fix lint

TarunRavikumar
authored and
charlesahn-scale
committed
use ecr image repo:tag directly

TarunRavikumar
authored and
charlesahn-scale
committed
fix: isort import ordering

TarunRavikumar
authored and
charlesahn-scale
committed
fix: remove unused infra_config import

TarunRavikumar
authored and
charlesahn-scale
committed
fix: mypy type annotation errors

TarunRavikumar
authored and
charlesahn-scale
committed
fix: remove type annotation causing mypy no-redef error

TarunRavikumar
authored and
charlesahn-scale
committed
fix: mypy type errors in s3_utils.py and io.py - use botocore.config.Config

TarunRavikumar
authored and
charlesahn-scale
committed
fix: mypy typeddict-item errors - use broad type ignore

TarunRavikumar
authored and
charlesahn-scale
committed
fix: update test mocks to use get_s3_resource from s3_utils

TarunRavikumar
authored and
charlesahn-scale
committed
test: add unit tests for s3_utils, onprem_docker_repository, and onprem_queue_endpoint_resource_delegate

TarunRavikumar
authored and
charlesahn-scale
committed
style: format test files with black

TarunRavikumar
authored and
charlesahn-scale
committed
refactor: use filesystem_gateway abstraction for S3 operations

TarunRavikumar
authored and
charlesahn-scale
committed
fix: deduplicate S3 client config by using centralized s3_utils

TarunRavikumar
authored and
charlesahn-scale
committed
fix: add pagination to list_objects to handle >1000 objects

TarunRavikumar
authored and
charlesahn-scale
committed
fix: make OnPremDockerRepository.get_image_url consistent with ECR/ACR

TarunRavikumar
authored and
charlesahn-scale
committed
refactor: add explicit on-prem branches in dependencies.py for clarity

TarunRavikumar
authored and
charlesahn-scale
committed
feat: implement Redis LLEN for queue depth in OnPremQueueEndpointResourceDelegate

TarunRavikumar
authored and
charlesahn-scale
committed
fix: replace mutable default argument with None in _get_client

TarunRavikumar
authored and
charlesahn-scale
committed
refactor: extract inline import to module-level helper function

TarunRavikumar
authored and
charlesahn-scale
committed
fix: reduce excessive debug logging in s3_utils

TarunRavikumar
authored and
charlesahn-scale
committed
chore: remove unused TYPE_CHECKING import

TarunRavikumar
authored and
charlesahn-scale
committed
fix: make Dockerfile multi-arch compatible for ARM/AMD64

TarunRavikumar
authored and
charlesahn-scale
committed
style: fix black formatting in test_onprem_queue_endpoint_resource_delegate

TarunRavikumar
authored and
charlesahn-scale
committed
fix: restore AWS_PROFILE env var fallback in s3_utils

TarunRavikumar
authored and
charlesahn-scale
committed
fix: correct isort ordering in s3_filesystem_gateway.py

TarunRavikumar
authored and
charlesahn-scale
committed
fix: use Literal type for s3 addressing_style to satisfy mypy

TarunRavikumar
authored and
charlesahn-scale
committed
Onprem Compatibility Change
charlesahn-scale
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Model Engine OnPrem Support and vLLM 0.11.1 + Model Engine Integration Fixes#744

Model Engine OnPrem Support and vLLM 0.11.1 + Model Engine Integration Fixes#744
charlesahn-scale wants to merge 30 commits intomainfrom
onPremSuport

Commits on Feb 3, 2026

add support for on-prem

clean up on-prem artificats

add back comments from initial code

fix lint

use ecr image repo:tag directly

fix: isort import ordering

fix: remove unused infra_config import

fix: mypy type annotation errors

fix: remove type annotation causing mypy no-redef error

fix: mypy type errors in s3_utils.py and io.py - use botocore.config.Config

fix: mypy typeddict-item errors - use broad type ignore

fix: update test mocks to use get_s3_resource from s3_utils

test: add unit tests for s3_utils, onprem_docker_repository, and onprem_queue_endpoint_resource_delegate

style: format test files with black

refactor: use filesystem_gateway abstraction for S3 operations

fix: deduplicate S3 client config by using centralized s3_utils

fix: add pagination to list_objects to handle >1000 objects

fix: make OnPremDockerRepository.get_image_url consistent with ECR/ACR

refactor: add explicit on-prem branches in dependencies.py for clarity

feat: implement Redis LLEN for queue depth in OnPremQueueEndpointResourceDelegate

fix: replace mutable default argument with None in _get_client

refactor: extract inline import to module-level helper function

fix: reduce excessive debug logging in s3_utils

chore: remove unused TYPE_CHECKING import

fix: make Dockerfile multi-arch compatible for ARM/AMD64

style: fix black formatting in test_onprem_queue_endpoint_resource_delegate

fix: restore AWS_PROFILE env var fallback in s3_utils

fix: correct isort ordering in s3_filesystem_gateway.py

fix: use Literal type for s3 addressing_style to satisfy mypy

Onprem Compatibility Change