Add multi-step tool-calling SDG tutorial for workplace assistant

[shashank3959]

Add notebook, tool definitions, and utility modules for generating synthetic multi-step tool-calling training data using Data Designer. Includes dual-level LLM judge filtering and NeMo Gym export.

[github-actions]

Thank you for your submission! We ask that you sign our Developer Certificate of Origin before we can accept your contribution. You can sign the DCO by adding a comment below using this text:

---

I have read the DCO document and I hereby sign the DCO.

---

_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the DCO Assistant Lite bot.}

[greptile-apps]

Greptile Overview

Greptile Summary

Adds comprehensive multi-step tool-calling synthetic data generation tutorial using Data Designer. Implements a complete pipeline for generating realistic workplace assistant queries with simulated agent trajectories, dual-level LLM judge filtering for quality control, and NeMo Gym export format compatibility.

Key Components:

• Tutorial notebook with 27 workplace assistant tools across 6 databases (email, calendar, CRM, analytics, project management, company directory)
• Dual-level quality filtering utilities to validate both user queries and generated trajectories
• NeMo Gym format conversion for RL training compatibility
• 27 multi-step patterns for diverse task generation (lookup-then-send, search-then-update, etc.)

Style Issue:

• All 3 Python utility files are missing required NVIDIA SPDX license headers as specified in AGENTS.md

Confidence Score: 4/5

• Safe to merge after adding license headers to Python files
• Well-structured tutorial with proper type annotations and good code organization. The only issue is missing NVIDIA license headers on 3 Python utility files, which is a style requirement that should be fixed before merging
• The 3 Python utility files in docs/colab_notebooks/5-multistep-toolcalling/utils/ need NVIDIA SPDX license headers added

Important Files Changed

Filename	Overview
docs/colab_notebooks/5-multistep-toolcalling/multistep-toolcalling.ipynb	Comprehensive tutorial notebook for multi-step tool-calling SDG with clear examples and dual-level quality filtering
docs/colab_notebooks/5-multistep-toolcalling/utils/init.py	Package initialization - missing NVIDIA license headers required per AGENTS.md
docs/colab_notebooks/5-multistep-toolcalling/utils/convert_to_nemo_gym_format.py	NeMo Gym format converter with proper type hints - missing NVIDIA license headers required per AGENTS.md
docs/colab_notebooks/5-multistep-toolcalling/utils/quality_filtering.py	Quality filtering utilities with dual-level validation - missing NVIDIA license headers required per AGENTS.md
docs/colab_notebooks/5-multistep-toolcalling/tools/environment.json	Environment configuration with 27 multi-step patterns covering all tool combinations

Sequence Diagram

sequenceDiagram
    participant User
    participant DataDesigner
    participant LLM
    participant QualityFilter
    participant NeMoGym

    User->>DataDesigner: Load tool schemas & seed data
    DataDesigner->>LLM: Generate user query from pattern
    LLM-->>DataDesigner: Return user query
    DataDesigner->>LLM: Judge user query (feasibility, schema compliance)
    LLM-->>DataDesigner: Return query scores
    DataDesigner->>LLM: Generate trajectory (tool calls)
    LLM-->>DataDesigner: Return agent trajectory
    DataDesigner->>LLM: Judge trajectory (tool validity, completeness)
    LLM-->>DataDesigner: Return trajectory scores
    DataDesigner->>QualityFilter: Filter by dual-level scores
    QualityFilter->>QualityFilter: Stage 1: Validate query
    QualityFilter->>QualityFilter: Stage 2: Validate trajectory
    QualityFilter-->>User: Return filtered dataset
    User->>NeMoGym: Convert to NeMo Gym JSONL format
    NeMoGym-->>User: Training data ready for RL

Loading

_{Last reviewed commit: eb2b52b}

[greptile-apps]

_{11 files reviewed, 3 comments}

_{Edit Code Review Agent Settings | Greptile}

[greptile-apps]

Missing NVIDIA license headers. Per AGENTS.md, all Python files must include:

Suggested change

	from .quality_filtering import filter_high_quality, show_rejection_reasons
	# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
	# SPDX-License-Identifier: Apache-2.0
	from .quality_filtering import filter_high_quality, show_rejection_reasons

Context Used: Context from dashboard - AGENTS.md (source)

Prompt To Fix With AI

This is a comment left during a code review.
Path: docs/colab_notebooks/5-multistep-toolcalling/utils/__init__.py
Line: 1:1

Comment:
Missing NVIDIA license headers. Per AGENTS.md, all Python files must include:

```suggestion
# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0

from .quality_filtering import filter_high_quality, show_rejection_reasons
```

**Context Used:** Context from `dashboard` - AGENTS.md ([source](https://app.greptile.com/review/custom-context?memory=95f8243f-5118-40bc-a3ab-69210b72e57e))

How can I resolve this? If you propose a fix, please make it concise.

[greptile-apps]

Missing NVIDIA license headers. Per AGENTS.md, all Python files must include:

Suggested change

	"""Utilities for converting generated records to NeMo Gym JSONL format."""
	# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
	# SPDX-License-Identifier: Apache-2.0
	"""Utilities for converting generated records to NeMo Gym JSONL format."""

Context Used: Context from dashboard - AGENTS.md (source)

Prompt To Fix With AI

This is a comment left during a code review.
Path: docs/colab_notebooks/5-multistep-toolcalling/utils/convert_to_nemo_gym_format.py
Line: 1:1

Comment:
Missing NVIDIA license headers. Per AGENTS.md, all Python files must include:

```suggestion
# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0

"""Utilities for converting generated records to NeMo Gym JSONL format."""
```

**Context Used:** Context from `dashboard` - AGENTS.md ([source](https://app.greptile.com/review/custom-context?memory=95f8243f-5118-40bc-a3ab-69210b72e57e))

How can I resolve this? If you propose a fix, please make it concise.

[greptile-apps]

Missing NVIDIA license headers. Per AGENTS.md, all Python files must include:

Suggested change

	"""Utilities for dual-level quality filtering of generated datasets."""
	# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
	# SPDX-License-Identifier: Apache-2.0
	"""Utilities for dual-level quality filtering of generated datasets."""

Context Used: Context from dashboard - AGENTS.md (source)

Prompt To Fix With AI

This is a comment left during a code review.
Path: docs/colab_notebooks/5-multistep-toolcalling/utils/quality_filtering.py
Line: 1:1

Comment:
Missing NVIDIA license headers. Per AGENTS.md, all Python files must include:

```suggestion
# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0

"""Utilities for dual-level quality filtering of generated datasets."""
```

**Context Used:** Context from `dashboard` - AGENTS.md ([source](https://app.greptile.com/review/custom-context?memory=95f8243f-5118-40bc-a3ab-69210b72e57e))

How can I resolve this? If you propose a fix, please make it concise.