{"id":30848328,"url":"https://github.com/dartayous/voice-intent-classifier","last_synced_at":"2026-05-14T20:33:02.762Z","repository":{"id":312353694,"uuid":"1047217107","full_name":"Dartayous/voice-intent-classifier","owner":"Dartayous","description":"“Modular ML pipeline for voice search intent classification with GPU profiling via Nsight Systems.”","archived":false,"fork":false,"pushed_at":"2025-08-30T01:19:40.000Z","size":2537,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2025-08-30T03:18:08.741Z","etag":null,"topics":["gpu","machine-learning","nlp","profiling","tensorflow"],"latest_commit_sha":null,"homepage":"","language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/Dartayous.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2025-08-30T00:17:01.000Z","updated_at":"2025-08-30T01:19:43.000Z","dependencies_parsed_at":"2025-08-30T03:18:12.067Z","dependency_job_id":"da6f76ec-fc22-4575-9da7-2ab31572f5d6","html_url":"https://github.com/Dartayous/voice-intent-classifier","commit_stats":null,"previous_names":["dartayous/voice-intent-classifier"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/Dartayous/voice-intent-classifier","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Dartayous%2Fvoice-intent-classifier","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Dartayous%2Fvoice-intent-classifier/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Dartayous%2Fvoice-intent-classifier/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Dartayous%2Fvoice-intent-classifier/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/Dartayous","download_url":"https://codeload.github.com/Dartayous/voice-intent-classifier/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/Dartayous%2Fvoice-intent-classifier/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":273990223,"owners_count":25203295,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-09-07T02:00:09.463Z","response_time":67,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["gpu","machine-learning","nlp","profiling","tensorflow"],"created_at":"2025-09-07T03:08:35.897Z","updated_at":"2026-05-14T20:33:02.756Z","avatar_url":"https://github.com/Dartayous.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"\u003cp align=\"center\"\u003e\n  \u003cimg src=\"assets/voice_intent_classifier_banner.png\" alt=\"Voice Intent Classifier\" width=\"100%\"\u003e\n\u003c/p\u003e\n\n#  Voice Search Intent Classifier (with GPU Profiling Lab)\n\nThis project trains an artificial neural network to classify voice search queries by intent—such as `information`, `command`, `shopping`, or `entertainment`. Built for clarity, modularity, and performance, the lab includes full GPU profiling using NVIDIA Nsight Systems and NVTX markers.\n\n---\n\n![CI](https://github.com/dartayous/ml-training-lab/actions/workflows/ci.yml/badge.svg)\n![CI](https://github.com/dartayous/voice-intent-classifier/actions/workflows/ci.yml/badge.svg)\n\n---\n\n##  Project Structure\n\nml-training-lab/ \n├── train_model.py    # Training script with NVTX profiling \n├── predict_intent.py # Inference script for new voice queries \n├── utils.py          # Shared preprocessing and encoding functions \n├── assets/           # .png files for viewing \n├── traces/           # Nsight Systems .nsys-rep files \n└── README.md         # Project overview and instructions\n\n\n---\n\n##  What This Project Demonstrates\n\n- ✅ Real-world NLP pipeline using voice search data\n- ✅ Intent classification with TensorFlow/Keras\n- ✅ Modular design for training, inference, and preprocessing\n- ✅ GPU profiling with Nsight Systems + NVTX annotations\n- ✅ Interview-ready explanations of model behavior and performance\n\n---\n\n##  Dataset\n\n- **Source**: `voice_search_query_captures.csv`\n- **Fields used**:\n  - `query_text`: Input to the model\n  - `intent`: Target label\n- **Preprocessing**:\n  - Lowercasing, punctuation removal\n  - Tokenization + padding\n  - Label encoding\n\n![Architecture Diagram](assets/Original_Data_Analyzed.png)\n\n---\n\n##  Model Architecture\n\n- `Embedding` layer (vocab size: 1000, dim: 16)\n- `GlobalAveragePooling1D`\n- `Dense` (24 units, ReLU)\n- `Dense` (softmax output for multi-class classification)\n\n![Architecture Diagram](assets/Trained_Model.png)\n![Architecture Diagram](assets/Tested_Model.png)\n\n---\n\n##  Architecture Overview\n\n![Architecture Diagram](assets/scene_1_architecture.png)\n![Architecture Diagram](assets/scene_2_architecture.png)\n![Architecture Diagram](assets/scene_3_architecture.png)\n![Architecture Diagram](assets/scene_4_architecture.png)\n![Architecture Diagram](assets/scene_5_architecture.png)\n![Architecture Diagram](assets/scene_6_architecture.png)\n\n--- \n\n##  Profiling with Nsight Systems\n\nTraining is wrapped with NVTX markers for each epoch:\n\n```python\nwith nvtx.annotate(\"Epoch 1\", color=\"blue\"):\n    model.fit(...)\n```\n---\n\n### To generate a trace:\nnsys profile -o traces/nn_training_trace --trace=cuda,nvtx python train_model.py\n\n\n### Open the .nsys-rep file in Nsight Systems to visualize:\n* CUDA kernel launches\n* Memory transfers\n* NVTX-labeled epochs\n* CPU orchestration\n\n![Architecture Diagram](assets/NVIDIA_Nsight_NLP_Interpretation_Training_Data.png)\n\n\n###  Inference Example\npython predict_intent.py\n\n### Sample output:\nQuery: 'Play jazz music on Spotify' → Predicted Intent: 'entertainment'\nQuery: 'Turn off the living room lights' → Predicted Intent: 'command'\n\n---\n\n##  Simulation Notebook\n\nExplore the full training and testing workflow in the interactive notebook:\n\n [`ml_training_lab_SIM.ipynb`](ml_training_lab_SIM.ipynb)\n\nThis notebook walks through:\n- Data loading and preprocessing\n- Model architecture and training\n- NVTX-wrapped profiling setup\n- Inference on new voice queries\n- Commentary and visual outputs\n\n---\n\n##  Requirements\n* Python 3.8+\n* TensorFlow\n* scikit-learn\n* NVIDIA Nsight Systems\n* NVTX (pip install nvtx)\n\n---\n\n##  Future Enhancements\n* Add confusion matrix and intent distribution visualizations\n* Integrate real-time voice input via microphone\n* Expand to multilingual intent classification\n* Profile inference latency and memory footprint\n\n---\n\n##  Author\nDartayous — Creative Technologist \u0026 AI/ML Engineer \nFocused on cinematic, modular GPU workflows and explainable AI. \nBuilt this lab to showcase real-world ML engineering and profiling mastery.\n\n---\n\n## 📎 License\nMIT License — feel free to fork, remix, and build upon it.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fdartayous%2Fvoice-intent-classifier","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fdartayous%2Fvoice-intent-classifier","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fdartayous%2Fvoice-intent-classifier/lists"}