fix: Major module mapping overhaul for auto-assign reviewers

- Replace broad 'CI/CD' with 5 granular modules
- Separate test concerns from pipeline logic
- Move module-owners.json to .github/ for consistency

Signed-off-by: Venky Ganesh <[email protected]>

Author: venkywonka

.github/module-owners.json

@@ -0,0 +1,39 @@
+{
+    "Generic Runtime": ["funatiq", "pcastonguay", "Shixiaowei02", "MartinMarciniszyn", "schetlur-nv", "dcampora"],
+    "Triton Backend": ["Tabrizian", "pcastonguay", "schetlur-nv"],
+    "LLM API/Workflow": ["Superjomn", "syuoni", "nv-guomingz", "litaotju", "QiJune"],
+    "KV-Cache Management":["thorjohnsen", "schetlur-nv"],
+    "Low Precision":["Tracin", "nv-guomingz", "Naveassaf"],
+    "Speculative Decoding":["yweng0828", "nekorobov", "lfr-0531"],
+    "Customized Kernels":["lowsfer", "PerkzZheng", "jdemouth-nvidia"],
+    "Performance": ["kaiyux", "jiahanc", "hypdeb"],
+    "Lora/P-tuning":["byshiue", "shaharmor98"],
+    "Disaggregated Serving":["Shixiaowei02", "joyang-nv", "chuangz0", "schetlur-nv"],
+    "Documentation":["nv-guomingz"],
+    "Sampling": ["dcampora", "lfr-0531", "Naveassaf", "syuoni", "yweng0828"],
+    "Memory": ["litaotju", "peaceh-nv"],
+    "Installation": ["hchings", "Superjomn", "nv-guomingz", "QiJune"],
+    "GitHub Configuration": ["tburt-nv", "niukuo"],
+    "Jenkins Pipelines": ["chzblych", "niukuo"],
+    "Test Configuration": ["niukuo", "syuoni", "LarryXFly"],
+    "Test Waive List": ["chzblych", "niukuo"],
+    "Integration Tests": ["LarryXFly", "niukuo"],
+    "Torch Framework": ["QiJune", "hlu1"],
+    "Torch Attention Backend": ["yuxianq", "hlu1"],
+    "Torch AutoDeploy": ["lucaslie", "suyoggupta"],
+    "Torch Compilation": ["litaotju", "yizhang-nv", "liji-nv"],
+    "Torch Custom Ops": ["yizhang-nv"],
+    "Torch Distributed": ["yilin-void", "yuxianq", "hyukn", "yizhang-nv", "hlu1"],
+    "Torch PyExecutor": ["dongxuy04", "funatiq", "dcampora", "HuiGao-NV"],
+    "Torch Speculative": ["lfr-0531", "mikeiovine"],
+    "Autotuner": ["hyukn", "litaotju"],
+    "Pipeline Interface": ["amukkara", "chang-l"],
+    "Torch Models": ["QiJune", "hlu1"],
+    "Torch Models DeepSeekV3": ["hlu1", "zongfeijing"],
+    "Torch Models Llama": ["chang-l", "mikeiovine"],
+    "Torch Modules": ["QiJune", "hlu1"],
+    "Torch Modules Attention": ["yuxianq", "hlu1"],
+    "Torch Modules Fused MOE": ["hlu1", "dongxuy04", "zongfeijing", "HuiGao-NV"],
+    "Torch Tests": ["QiJune", "hlu1"],
+    "PyTorch Examples": ["QiJune", "hlu1"]
+}

.github/module-paths.json

@@ -7,8 +7,11 @@
     "examples/disaggregated/": "Disaggregated Serving",
     "docs/": "Documentation",
     "docker/": "Installation",
-    ".github/": "CI/CD",
-    "jenkins/": "CI/CD",
+    ".github/": "GitHub Configuration",
+    "jenkins/": "Jenkins Pipelines",
+    "tests/integration/test_lists/": "Test Configuration",
+    "tests/integration/test_lists/waives.txt": "Test Waive List",
+    "tests/integration/defs/": "Integration Tests",
     "tensorrt_llm/_torch/": "Torch Framework",
     "tensorrt_llm/_torch/attention_backend/": "Torch Attention Backend",
     "tensorrt_llm/_torch/auto_deploy/": "Torch AutoDeploy",

.github/scripts/assign_reviewers.py

@@ -137,8 +137,7 @@ def main() -> None:
         print(f"Changed files: {changed_files}")
 
         module_paths = load_json(Path(".github") / "module-paths.json")
-        module_owners = load_json(
-            Path(".github/workflows") / "module-owners.json")
+        module_owners = load_json(Path(".github") / "module-owners.json")
 
         modules = map_modules(changed_files, module_paths)
         reviewers = gather_reviewers(

CONTRIBUTING.md

@@ -93,13 +93,16 @@ Developer workflow for code contributions is as follows:
 3. Once the code changes are staged on the fork and ready for review, a [Pull Request](https://help.github.com/en/articles/about-pull-requests) (PR) can be [requested](https://help.github.com/en/articles/creating-a-pull-request) to merge the changes from a branch of the fork into a selected branch of upstream. PRs should typically target the `main` branch.
   * Creation of a PR creation kicks off the code review process.
   * At least one TensorRT-LLM engineer will be assigned for the review. When the PR is under review, the label `Pending Review` will be added to the PR.
+  * If changes are requested, then the reviewer will add the label `Changes Requested` to the PR.
+  * Once changes are approved, CI will be launched to validate the change. When CI passes, the reviewer will merge the PR.
+  * If CI reports any failures, it's up to the requester to fix any CI failures before requesting another review.
 
 ### Automatic Reviewer Assignment
 
 Reviewers are automatically assigned to PRs through a GitHub Action that:
 
 * **Triggers**: Runs automatically when PRs are opened, synchronized, or reopened
-* **Module-based assignment**: Maps changed files to modules using `.github/module-paths.json` and assigns reviewers based on module ownership defined in `.github/workflows/module-owners.json`
+* **Module-based assignment**: Maps changed files to modules using `.github/module-paths.json` and assigns reviewers based on module ownership defined in `.github/module-owners.json`
 * **Respects existing assignments**: Won't assign additional reviewers if any reviewers are already assigned (unless forced)
 * **Excludes PR author**: Automatically excludes the PR author from reviewer assignments
 * **Limits reviewer count**: Randomly samples up to 3 reviewers if more are eligible (configurable via `REVIEWER_LIMIT`)
@@ -114,9 +117,6 @@ The auto-assignment system analyzes all files changed in your PR, maps them to t
   ```
 
 **Manual assignment**: You can also manually trigger reviewer assignment via GitHub's workflow dispatch with options for dry-run mode and force-assignment.
-  * If changes are requested, then the reviewer will add the label `Changes Requested` to the PR.
-  * Once changes are approved, CI will be launched to validate the change. When CI passes, the reviewer will merge the PR.
-  * If CI reports any failures, it's up to the requester to fix any CI failures before requesting another review.
 
 ### PR Submission Policies