Release 3.17.0

[mohamedzeidan2021]

Bump versions and dependency pins, update CHANGELOGs for the 3.17.0 release (sagemaker-core 2.17.0, sagemaker-train/serve/mlops 1.17.0).

Anchor: Release 3.16.0 (3f810c6). Cutoff: d8ded10.

Issue #, if available:

Description of changes:

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[github-actions]

Claude finished @mohamedzeidan2021's task in 47s —— View job

---

Release 3.17.0 Review

• Read PR diff (/tmp/pr.diff)
• Verify version bumps across all packages
• Verify dependency pin consistency
• Cross-check CHANGELOG entries vs. sub-package changelogs

---

This is a clean release PR. I verified the mechanical correctness of the version and dependency changes:

Version bumps — consistent ✅

Package	Old → New
sagemaker	3.16.0 → 3.17.0
sagemaker-core	2.16.0 → 2.17.0
sagemaker-train / serve / mlops	1.16.0 → 1.17.0

Each VERSION file matches its ## vX.Y.Z (2026-07-24) CHANGELOG heading, and today's date (2026-07-24) is correct.

Dependency pins — consistent ✅
All intra-package pins were bumped in lockstep with the new floors, and the dependency hierarchy holds:

• root pyproject.toml: core>=2.17.0, train/serve/mlops>=1.17.0
• sagemaker-mlops: core>=2.17.0, train>=1.17.0, serve>=1.17.0
• sagemaker-serve: core>=2.17.0, train>=1.17.0
• sagemaker-train: core>=2.17.0

No stale 1.16.0/2.16.0 references remain, and no lower/upper bound was accidentally dropped.

CHANGELOG cross-check — consistent ✅
Every PR listed in a sub-package changelog also appears in the root aggregate:

• core (feat: update SDK to use latest LMI v27 image for sdk v3.x #5976, Fix source_dir in FrameworkProcessor #6047), serve (feat(serve): support fine-tuned models in deployment-config API #6041, Fix private hub #6036, fix(serve): support aliased hub content names in private hub deploys #6039, fix(serve): dedicated INFERENCE_RECOMMENDER telemetry feature + type workload param #6028, fix: Fixing EULA check, relying on HostingEulaUri field #6077, test: move two tests in serve to gpu-integ-tests #6096), train (fix(train): correct Networking field names in ModelTrainer intelligen… #6064, fix: resolve MTRL eval base-model ARN against the configured hub (not always SageMakerPublicHub) #6040, fix(train): Fall back to public hub when private hub lacks base model #6092, datamixing recipe path fix #6073, Doc update and added SFT integ test #6018, Fix role issue in mtrl integ tests #6070, Fix gpu integ test failure due to outdated MPG #6097), mlops (feat: Wire BatchWriteRecord and ListRecords into ingest_dataframe #6026, test(feature-processor): Isolate pipeline names to fix flaky integ tests #6095) — all present in root.
• Shared test entries test(integ): absorb iam:SimulatePrincipalPolicy throttling across integ suites #6081 and test(integ): let exhausted IAM throttling fail instead of skipping #6094 are correctly duplicated across the relevant packages.

Minor observations (non-blocking, no change requested):

• I can only validate the changelog against the diff itself — I can't confirm the anchor→cutoff range (3f810c6…d8ded10f7) captured every merged PR, since that requires the git log. Worth a quick manual sanity check that nothing merged in-window was omitted.
• #6047 (Fix source_dir in FrameworkProcessor) is filed under sagemaker-core — just confirm that fix actually landed in the core package and not elsewhere.

No correctness, security, or backward-compatibility concerns. LGTM. 👍

[mohamedzeidan2021]

Bump versions and dependency pins, update CHANGELOGs for the 3.17.0 release (sagemaker-core 2.17.0, sagemaker-train/serve/mlops 1.17.0).

Anchor: Release 3.16.0 (3f810c6). Cutoff: d8ded10.

Issue #, if available:

Description of changes:

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[github-actions]

Claude finished @mohamedzeidan2021's task in 47s —— View job

---

Release 3.17.0 Review

• Read PR diff (/tmp/pr.diff)
• Verify version bumps across all packages
• Verify dependency pin consistency
• Cross-check CHANGELOG entries vs. sub-package changelogs

---

This is a clean release PR. I verified the mechanical correctness of the version and dependency changes:

Version bumps — consistent ✅

Package	Old → New
sagemaker	3.16.0 → 3.17.0
sagemaker-core	2.16.0 → 2.17.0
sagemaker-train / serve / mlops	1.16.0 → 1.17.0

Each VERSION file matches its ## vX.Y.Z (2026-07-24) CHANGELOG heading, and today's date (2026-07-24) is correct.

Dependency pins — consistent ✅
All intra-package pins were bumped in lockstep with the new floors, and the dependency hierarchy holds:

• root pyproject.toml: core>=2.17.0, train/serve/mlops>=1.17.0
• sagemaker-mlops: core>=2.17.0, train>=1.17.0, serve>=1.17.0
• sagemaker-serve: core>=2.17.0, train>=1.17.0
• sagemaker-train: core>=2.17.0

No stale 1.16.0/2.16.0 references remain, and no lower/upper bound was accidentally dropped.

CHANGELOG cross-check — consistent ✅
Every PR listed in a sub-package changelog also appears in the root aggregate:

• core (feat: update SDK to use latest LMI v27 image for sdk v3.x #5976, Fix source_dir in FrameworkProcessor #6047), serve (feat(serve): support fine-tuned models in deployment-config API #6041, Fix private hub #6036, fix(serve): support aliased hub content names in private hub deploys #6039, fix(serve): dedicated INFERENCE_RECOMMENDER telemetry feature + type workload param #6028, fix: Fixing EULA check, relying on HostingEulaUri field #6077, test: move two tests in serve to gpu-integ-tests #6096), train (fix(train): correct Networking field names in ModelTrainer intelligen… #6064, fix: resolve MTRL eval base-model ARN against the configured hub (not always SageMakerPublicHub) #6040, fix(train): Fall back to public hub when private hub lacks base model #6092, datamixing recipe path fix #6073, Doc update and added SFT integ test #6018, Fix role issue in mtrl integ tests #6070, Fix gpu integ test failure due to outdated MPG #6097), mlops (feat: Wire BatchWriteRecord and ListRecords into ingest_dataframe #6026, test(feature-processor): Isolate pipeline names to fix flaky integ tests #6095) — all present in root.
• Shared test entries test(integ): absorb iam:SimulatePrincipalPolicy throttling across integ suites #6081 and test(integ): let exhausted IAM throttling fail instead of skipping #6094 are correctly duplicated across the relevant packages.

Minor observations (non-blocking, no change requested):

• I can only validate the changelog against the diff itself — I can't confirm the anchor→cutoff range (3f810c6…d8ded10f7) captured every merged PR, since that requires the git log. Worth a quick manual sanity check that nothing merged in-window was omitted.
• #6047 (Fix source_dir in FrameworkProcessor) is filed under sagemaker-core — just confirm that fix actually landed in the core package and not elsewhere.

No correctness, security, or backward-compatibility concerns. LGTM. 👍

[mohamedzeidan2021]

Bump versions and dependency pins, update CHANGELOGs for the 3.17.0 release (sagemaker-core 2.17.0, sagemaker-train/serve/mlops 1.17.0).

Anchor: Release 3.16.0 (3f810c6). Cutoff: d8ded10.

Issue #, if available:

Description of changes:

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[github-actions]

Claude finished @mohamedzeidan2021's task in 47s —— View job

---

Release 3.17.0 Review

• Read PR diff (/tmp/pr.diff)
• Verify version bumps across all packages
• Verify dependency pin consistency
• Cross-check CHANGELOG entries vs. sub-package changelogs

---

This is a clean release PR. I verified the mechanical correctness of the version and dependency changes:

Version bumps — consistent ✅

Package	Old → New
sagemaker	3.16.0 → 3.17.0
sagemaker-core	2.16.0 → 2.17.0
sagemaker-train / serve / mlops	1.16.0 → 1.17.0

Each VERSION file matches its ## vX.Y.Z (2026-07-24) CHANGELOG heading, and today's date (2026-07-24) is correct.

Dependency pins — consistent ✅
All intra-package pins were bumped in lockstep with the new floors, and the dependency hierarchy holds:

• root pyproject.toml: core>=2.17.0, train/serve/mlops>=1.17.0
• sagemaker-mlops: core>=2.17.0, train>=1.17.0, serve>=1.17.0
• sagemaker-serve: core>=2.17.0, train>=1.17.0
• sagemaker-train: core>=2.17.0

No stale 1.16.0/2.16.0 references remain, and no lower/upper bound was accidentally dropped.

CHANGELOG cross-check — consistent ✅
Every PR listed in a sub-package changelog also appears in the root aggregate:

• core (feat: update SDK to use latest LMI v27 image for sdk v3.x #5976, Fix source_dir in FrameworkProcessor #6047), serve (feat(serve): support fine-tuned models in deployment-config API #6041, Fix private hub #6036, fix(serve): support aliased hub content names in private hub deploys #6039, fix(serve): dedicated INFERENCE_RECOMMENDER telemetry feature + type workload param #6028, fix: Fixing EULA check, relying on HostingEulaUri field #6077, test: move two tests in serve to gpu-integ-tests #6096), train (fix(train): correct Networking field names in ModelTrainer intelligen… #6064, fix: resolve MTRL eval base-model ARN against the configured hub (not always SageMakerPublicHub) #6040, fix(train): Fall back to public hub when private hub lacks base model #6092, datamixing recipe path fix #6073, Doc update and added SFT integ test #6018, Fix role issue in mtrl integ tests #6070, Fix gpu integ test failure due to outdated MPG #6097), mlops (feat: Wire BatchWriteRecord and ListRecords into ingest_dataframe #6026, test(feature-processor): Isolate pipeline names to fix flaky integ tests #6095) — all present in root.
• Shared test entries test(integ): absorb iam:SimulatePrincipalPolicy throttling across integ suites #6081 and test(integ): let exhausted IAM throttling fail instead of skipping #6094 are correctly duplicated across the relevant packages.

Minor observations (non-blocking, no change requested):

• I can only validate the changelog against the diff itself — I can't confirm the anchor→cutoff range (3f810c6…d8ded10f7) captured every merged PR, since that requires the git log. Worth a quick manual sanity check that nothing merged in-window was omitted.
• #6047 (Fix source_dir in FrameworkProcessor) is filed under sagemaker-core — just confirm that fix actually landed in the core package and not elsewhere.

No correctness, security, or backward-compatibility concerns. LGTM. 👍

[mohamedzeidan2021]

Bump versions and dependency pins, update CHANGELOGs for the 3.17.0 release (sagemaker-core 2.17.0, sagemaker-train/serve/mlops 1.17.0).

Anchor: Release 3.16.0 (3f810c6). Cutoff: d8ded10.

Issue #, if available:

Description of changes:

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[github-actions]

Claude finished @mohamedzeidan2021's task in 47s —— View job

---

Release 3.17.0 Review

• Read PR diff (/tmp/pr.diff)
• Verify version bumps across all packages
• Verify dependency pin consistency
• Cross-check CHANGELOG entries vs. sub-package changelogs

---

This is a clean release PR. I verified the mechanical correctness of the version and dependency changes:

Version bumps — consistent ✅

Package	Old → New
sagemaker	3.16.0 → 3.17.0
sagemaker-core	2.16.0 → 2.17.0
sagemaker-train / serve / mlops	1.16.0 → 1.17.0

Each VERSION file matches its ## vX.Y.Z (2026-07-24) CHANGELOG heading, and today's date (2026-07-24) is correct.

Dependency pins — consistent ✅
All intra-package pins were bumped in lockstep with the new floors, and the dependency hierarchy holds:

• root pyproject.toml: core>=2.17.0, train/serve/mlops>=1.17.0
• sagemaker-mlops: core>=2.17.0, train>=1.17.0, serve>=1.17.0
• sagemaker-serve: core>=2.17.0, train>=1.17.0
• sagemaker-train: core>=2.17.0

No stale 1.16.0/2.16.0 references remain, and no lower/upper bound was accidentally dropped.

CHANGELOG cross-check — consistent ✅
Every PR listed in a sub-package changelog also appears in the root aggregate:

• core (feat: update SDK to use latest LMI v27 image for sdk v3.x #5976, Fix source_dir in FrameworkProcessor #6047), serve (feat(serve): support fine-tuned models in deployment-config API #6041, Fix private hub #6036, fix(serve): support aliased hub content names in private hub deploys #6039, fix(serve): dedicated INFERENCE_RECOMMENDER telemetry feature + type workload param #6028, fix: Fixing EULA check, relying on HostingEulaUri field #6077, test: move two tests in serve to gpu-integ-tests #6096), train (fix(train): correct Networking field names in ModelTrainer intelligen… #6064, fix: resolve MTRL eval base-model ARN against the configured hub (not always SageMakerPublicHub) #6040, fix(train): Fall back to public hub when private hub lacks base model #6092, datamixing recipe path fix #6073, Doc update and added SFT integ test #6018, Fix role issue in mtrl integ tests #6070, Fix gpu integ test failure due to outdated MPG #6097), mlops (feat: Wire BatchWriteRecord and ListRecords into ingest_dataframe #6026, test(feature-processor): Isolate pipeline names to fix flaky integ tests #6095) — all present in root.
• Shared test entries test(integ): absorb iam:SimulatePrincipalPolicy throttling across integ suites #6081 and test(integ): let exhausted IAM throttling fail instead of skipping #6094 are correctly duplicated across the relevant packages.

Minor observations (non-blocking, no change requested):

• I can only validate the changelog against the diff itself — I can't confirm the anchor→cutoff range (3f810c6…d8ded10f7) captured every merged PR, since that requires the git log. Worth a quick manual sanity check that nothing merged in-window was omitted.
• #6047 (Fix source_dir in FrameworkProcessor) is filed under sagemaker-core — just confirm that fix actually landed in the core package and not elsewhere.

No correctness, security, or backward-compatibility concerns. LGTM. 👍

[mohamedzeidan2021]

Bump versions and dependency pins, update CHANGELOGs for the 3.17.0 release (sagemaker-core 2.17.0, sagemaker-train/serve/mlops 1.17.0).

Anchor: Release 3.16.0 (3f810c6). Cutoff: d8ded10.

Issue #, if available:

Description of changes:

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[github-actions]

Claude finished @mohamedzeidan2021's task in 47s —— View job

---

Release 3.17.0 Review

• Read PR diff (/tmp/pr.diff)
• Verify version bumps across all packages
• Verify dependency pin consistency
• Cross-check CHANGELOG entries vs. sub-package changelogs

---

This is a clean release PR. I verified the mechanical correctness of the version and dependency changes:

Version bumps — consistent ✅

Package	Old → New
sagemaker	3.16.0 → 3.17.0
sagemaker-core	2.16.0 → 2.17.0
sagemaker-train / serve / mlops	1.16.0 → 1.17.0

Each VERSION file matches its ## vX.Y.Z (2026-07-24) CHANGELOG heading, and today's date (2026-07-24) is correct.

Dependency pins — consistent ✅
All intra-package pins were bumped in lockstep with the new floors, and the dependency hierarchy holds:

• root pyproject.toml: core>=2.17.0, train/serve/mlops>=1.17.0
• sagemaker-mlops: core>=2.17.0, train>=1.17.0, serve>=1.17.0
• sagemaker-serve: core>=2.17.0, train>=1.17.0
• sagemaker-train: core>=2.17.0

No stale 1.16.0/2.16.0 references remain, and no lower/upper bound was accidentally dropped.

CHANGELOG cross-check — consistent ✅
Every PR listed in a sub-package changelog also appears in the root aggregate:

• core (feat: update SDK to use latest LMI v27 image for sdk v3.x #5976, Fix source_dir in FrameworkProcessor #6047), serve (feat(serve): support fine-tuned models in deployment-config API #6041, Fix private hub #6036, fix(serve): support aliased hub content names in private hub deploys #6039, fix(serve): dedicated INFERENCE_RECOMMENDER telemetry feature + type workload param #6028, fix: Fixing EULA check, relying on HostingEulaUri field #6077, test: move two tests in serve to gpu-integ-tests #6096), train (fix(train): correct Networking field names in ModelTrainer intelligen… #6064, fix: resolve MTRL eval base-model ARN against the configured hub (not always SageMakerPublicHub) #6040, fix(train): Fall back to public hub when private hub lacks base model #6092, datamixing recipe path fix #6073, Doc update and added SFT integ test #6018, Fix role issue in mtrl integ tests #6070, Fix gpu integ test failure due to outdated MPG #6097), mlops (feat: Wire BatchWriteRecord and ListRecords into ingest_dataframe #6026, test(feature-processor): Isolate pipeline names to fix flaky integ tests #6095) — all present in root.
• Shared test entries test(integ): absorb iam:SimulatePrincipalPolicy throttling across integ suites #6081 and test(integ): let exhausted IAM throttling fail instead of skipping #6094 are correctly duplicated across the relevant packages.

Minor observations (non-blocking, no change requested):

• I can only validate the changelog against the diff itself — I can't confirm the anchor→cutoff range (3f810c6…d8ded10f7) captured every merged PR, since that requires the git log. Worth a quick manual sanity check that nothing merged in-window was omitted.
• #6047 (Fix source_dir in FrameworkProcessor) is filed under sagemaker-core — just confirm that fix actually landed in the core package and not elsewhere.

No correctness, security, or backward-compatibility concerns. LGTM. 👍

[mohamedzeidan2021]

Bump versions and dependency pins, update CHANGELOGs for the 3.17.0 release (sagemaker-core 2.17.0, sagemaker-train/serve/mlops 1.17.0).

Anchor: Release 3.16.0 (3f810c6). Cutoff: d8ded10.

Issue #, if available:

Description of changes:

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[github-actions]

Claude finished @mohamedzeidan2021's task in 47s —— View job

---

Release 3.17.0 Review

• Read PR diff (/tmp/pr.diff)
• Verify version bumps across all packages
• Verify dependency pin consistency
• Cross-check CHANGELOG entries vs. sub-package changelogs

---

This is a clean release PR. I verified the mechanical correctness of the version and dependency changes:

Version bumps — consistent ✅

Package	Old → New
sagemaker	3.16.0 → 3.17.0
sagemaker-core	2.16.0 → 2.17.0
sagemaker-train / serve / mlops	1.16.0 → 1.17.0

Each VERSION file matches its ## vX.Y.Z (2026-07-24) CHANGELOG heading, and today's date (2026-07-24) is correct.

Dependency pins — consistent ✅
All intra-package pins were bumped in lockstep with the new floors, and the dependency hierarchy holds:

• root pyproject.toml: core>=2.17.0, train/serve/mlops>=1.17.0
• sagemaker-mlops: core>=2.17.0, train>=1.17.0, serve>=1.17.0
• sagemaker-serve: core>=2.17.0, train>=1.17.0
• sagemaker-train: core>=2.17.0

No stale 1.16.0/2.16.0 references remain, and no lower/upper bound was accidentally dropped.

CHANGELOG cross-check — consistent ✅
Every PR listed in a sub-package changelog also appears in the root aggregate:

• core (feat: update SDK to use latest LMI v27 image for sdk v3.x #5976, Fix source_dir in FrameworkProcessor #6047), serve (feat(serve): support fine-tuned models in deployment-config API #6041, Fix private hub #6036, fix(serve): support aliased hub content names in private hub deploys #6039, fix(serve): dedicated INFERENCE_RECOMMENDER telemetry feature + type workload param #6028, fix: Fixing EULA check, relying on HostingEulaUri field #6077, test: move two tests in serve to gpu-integ-tests #6096), train (fix(train): correct Networking field names in ModelTrainer intelligen… #6064, fix: resolve MTRL eval base-model ARN against the configured hub (not always SageMakerPublicHub) #6040, fix(train): Fall back to public hub when private hub lacks base model #6092, datamixing recipe path fix #6073, Doc update and added SFT integ test #6018, Fix role issue in mtrl integ tests #6070, Fix gpu integ test failure due to outdated MPG #6097), mlops (feat: Wire BatchWriteRecord and ListRecords into ingest_dataframe #6026, test(feature-processor): Isolate pipeline names to fix flaky integ tests #6095) — all present in root.
• Shared test entries test(integ): absorb iam:SimulatePrincipalPolicy throttling across integ suites #6081 and test(integ): let exhausted IAM throttling fail instead of skipping #6094 are correctly duplicated across the relevant packages.

Minor observations (non-blocking, no change requested):

• I can only validate the changelog against the diff itself — I can't confirm the anchor→cutoff range (3f810c6…d8ded10f7) captured every merged PR, since that requires the git log. Worth a quick manual sanity check that nothing merged in-window was omitted.
• #6047 (Fix source_dir in FrameworkProcessor) is filed under sagemaker-core — just confirm that fix actually landed in the core package and not elsewhere.

No correctness, security, or backward-compatibility concerns. LGTM. 👍

[mohamedzeidan2021]

Bump versions and dependency pins, update CHANGELOGs for the 3.17.0 release (sagemaker-core 2.17.0, sagemaker-train/serve/mlops 1.17.0).

Anchor: Release 3.16.0 (3f810c6). Cutoff: d8ded10.

Issue #, if available:

Description of changes:

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[github-actions]

Claude finished @mohamedzeidan2021's task in 47s —— View job

---

Release 3.17.0 Review

• Read PR diff (/tmp/pr.diff)
• Verify version bumps across all packages
• Verify dependency pin consistency
• Cross-check CHANGELOG entries vs. sub-package changelogs

---

This is a clean release PR. I verified the mechanical correctness of the version and dependency changes:

Version bumps — consistent ✅

Package	Old → New
sagemaker	3.16.0 → 3.17.0
sagemaker-core	2.16.0 → 2.17.0
sagemaker-train / serve / mlops	1.16.0 → 1.17.0

Each VERSION file matches its ## vX.Y.Z (2026-07-24) CHANGELOG heading, and today's date (2026-07-24) is correct.

Dependency pins — consistent ✅
All intra-package pins were bumped in lockstep with the new floors, and the dependency hierarchy holds:

• root pyproject.toml: core>=2.17.0, train/serve/mlops>=1.17.0
• sagemaker-mlops: core>=2.17.0, train>=1.17.0, serve>=1.17.0
• sagemaker-serve: core>=2.17.0, train>=1.17.0
• sagemaker-train: core>=2.17.0

No stale 1.16.0/2.16.0 references remain, and no lower/upper bound was accidentally dropped.

CHANGELOG cross-check — consistent ✅
Every PR listed in a sub-package changelog also appears in the root aggregate:

• core (feat: update SDK to use latest LMI v27 image for sdk v3.x #5976, Fix source_dir in FrameworkProcessor #6047), serve (feat(serve): support fine-tuned models in deployment-config API #6041, Fix private hub #6036, fix(serve): support aliased hub content names in private hub deploys #6039, fix(serve): dedicated INFERENCE_RECOMMENDER telemetry feature + type workload param #6028, fix: Fixing EULA check, relying on HostingEulaUri field #6077, test: move two tests in serve to gpu-integ-tests #6096), train (fix(train): correct Networking field names in ModelTrainer intelligen… #6064, fix: resolve MTRL eval base-model ARN against the configured hub (not always SageMakerPublicHub) #6040, fix(train): Fall back to public hub when private hub lacks base model #6092, datamixing recipe path fix #6073, Doc update and added SFT integ test #6018, Fix role issue in mtrl integ tests #6070, Fix gpu integ test failure due to outdated MPG #6097), mlops (feat: Wire BatchWriteRecord and ListRecords into ingest_dataframe #6026, test(feature-processor): Isolate pipeline names to fix flaky integ tests #6095) — all present in root.
• Shared test entries test(integ): absorb iam:SimulatePrincipalPolicy throttling across integ suites #6081 and test(integ): let exhausted IAM throttling fail instead of skipping #6094 are correctly duplicated across the relevant packages.

Minor observations (non-blocking, no change requested):

• I can only validate the changelog against the diff itself — I can't confirm the anchor→cutoff range (3f810c6…d8ded10f7) captured every merged PR, since that requires the git log. Worth a quick manual sanity check that nothing merged in-window was omitted.
• #6047 (Fix source_dir in FrameworkProcessor) is filed under sagemaker-core — just confirm that fix actually landed in the core package and not elsewhere.

No correctness, security, or backward-compatibility concerns. LGTM. 👍

[mohamedzeidan2021]

Bump versions and dependency pins, update CHANGELOGs for the 3.17.0 release (sagemaker-core 2.17.0, sagemaker-train/serve/mlops 1.17.0).

Anchor: Release 3.16.0 (3f810c6). Cutoff: d8ded10.

Issue #, if available:

Description of changes:

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[github-actions]

Claude finished @mohamedzeidan2021's task in 47s —— View job

---

Release 3.17.0 Review

• Read PR diff (/tmp/pr.diff)
• Verify version bumps across all packages
• Verify dependency pin consistency
• Cross-check CHANGELOG entries vs. sub-package changelogs

---

This is a clean release PR. I verified the mechanical correctness of the version and dependency changes:

Version bumps — consistent ✅

Package	Old → New
sagemaker	3.16.0 → 3.17.0
sagemaker-core	2.16.0 → 2.17.0
sagemaker-train / serve / mlops	1.16.0 → 1.17.0

Each VERSION file matches its ## vX.Y.Z (2026-07-24) CHANGELOG heading, and today's date (2026-07-24) is correct.

Dependency pins — consistent ✅
All intra-package pins were bumped in lockstep with the new floors, and the dependency hierarchy holds:

• root pyproject.toml: core>=2.17.0, train/serve/mlops>=1.17.0
• sagemaker-mlops: core>=2.17.0, train>=1.17.0, serve>=1.17.0
• sagemaker-serve: core>=2.17.0, train>=1.17.0
• sagemaker-train: core>=2.17.0

No stale 1.16.0/2.16.0 references remain, and no lower/upper bound was accidentally dropped.

CHANGELOG cross-check — consistent ✅
Every PR listed in a sub-package changelog also appears in the root aggregate:

• core (feat: update SDK to use latest LMI v27 image for sdk v3.x #5976, Fix source_dir in FrameworkProcessor #6047), serve (feat(serve): support fine-tuned models in deployment-config API #6041, Fix private hub #6036, fix(serve): support aliased hub content names in private hub deploys #6039, fix(serve): dedicated INFERENCE_RECOMMENDER telemetry feature + type workload param #6028, fix: Fixing EULA check, relying on HostingEulaUri field #6077, test: move two tests in serve to gpu-integ-tests #6096), train (fix(train): correct Networking field names in ModelTrainer intelligen… #6064, fix: resolve MTRL eval base-model ARN against the configured hub (not always SageMakerPublicHub) #6040, fix(train): Fall back to public hub when private hub lacks base model #6092, datamixing recipe path fix #6073, Doc update and added SFT integ test #6018, Fix role issue in mtrl integ tests #6070, Fix gpu integ test failure due to outdated MPG #6097), mlops (feat: Wire BatchWriteRecord and ListRecords into ingest_dataframe #6026, test(feature-processor): Isolate pipeline names to fix flaky integ tests #6095) — all present in root.
• Shared test entries test(integ): absorb iam:SimulatePrincipalPolicy throttling across integ suites #6081 and test(integ): let exhausted IAM throttling fail instead of skipping #6094 are correctly duplicated across the relevant packages.

Minor observations (non-blocking, no change requested):

• I can only validate the changelog against the diff itself — I can't confirm the anchor→cutoff range (3f810c6…d8ded10f7) captured every merged PR, since that requires the git log. Worth a quick manual sanity check that nothing merged in-window was omitted.
• #6047 (Fix source_dir in FrameworkProcessor) is filed under sagemaker-core — just confirm that fix actually landed in the core package and not elsewhere.

No correctness, security, or backward-compatibility concerns. LGTM. 👍