A curated collection of machine-learning projects, organized as Git submodules. Each submodule is an independent, self-contained repository — this hub provides a single entry point for cloning, browsing, and keeping them up to date.
| Project | Domain | Stack | Description |
|---|---|---|---|
| CLASSMATE-RAG | NLP / RAG | Python, ChromaDB, llama.cpp | Local-first retrieval-augmented generation system for course materials. Hybrid BM25 + vector retrieval with subject-aware routing, multilingual embeddings, and offline GGUF LLMs. |
| Speech-Digit-Classification | Audio / CV | TensorFlow, CNN | Spoken-digit (0–9) classifier trained on Mel-spectrogram features from the Free Spoken Digit Dataset. |
| Health-Risk-Predictor | Healthcare / DL | TensorFlow | Clinical survival prediction using a deep neural network. |
| Tweet-Sentiment-Analysis | NLP | Python | Predicting emotional polarity in tweets. |
Clone with all submodules:
git clone --recurse-submodules https://github.com/taha-kms/ML-Projects-Hub.gitIf you've already cloned without submodules:
make initA Makefile wraps the common submodule chores:
make init # initialize submodules after a plain clone
make update # pull the latest commit on main for every submodule
make status # show submodule status
make foreach CMD="git log -1 --oneline" # run a command in each submoduleAll submodules track their main branch (configured in .gitmodules), so make update advances each one to the latest upstream commit and stages the new pointers in this repo.
ML-Projects-Hub/
├── CLASSMATE-RAG/ (submodule)
├── Health-Risk-Predictor/ (submodule)
├── Speech-Digit-Classification/ (submodule)
├── Tweet-Sentiment-Analysis/ (submodule)
├── .github/workflows/ CI to keep submodules fresh
├── CONTRIBUTING.md
├── LICENSE
├── Makefile
└── README.md
See CONTRIBUTING.md. New ML projects, fixes, and documentation improvements are welcome.
The hub itself is released under the MIT License. Each linked submodule is governed by its own license — check the individual repository.