A curated collection of end-to-end Artificial Intelligence and Machine Learning projects demonstrating the complete lifecycle from data preparation and model training to production deployment on cloud platforms.
A production-grade 4-class brain tumor MRI classification system deployed on Google Cloud Platform.
- Model: VGG16 (Transfer Learning) — 94.44% test accuracy
- Classes: Glioma, Meningioma, No Tumor, Pituitary
- Backend: FastAPI + Docker + Cloud Run
- Frontend: Next.js + Tailwind CSS + Cloud Run
- MLOps: Vertex AI Model Registry + Cloud Build CI/CD
- Training: Two-phase fine-tuning on Kaggle Tesla T4 GPU
- Repository: https://github.com/prem332/brain-tumor-image-classification
The live deployment URL may be temporarily suspended to avoid ongoing GCP billing charges. The project is fully functional and can be redeployed within 15–20 minutes using the CI/CD pipeline. Production screenshots are available in the repository as proof of the working deployment.
A production-grade mental health sentiment classification system deployed on Google Cloud Platform.
The live deployment URL may be temporarily suspended to avoid ongoing GCP billing charges.
| Project | Type | Tech Stack |
|---|---|---|
| Brain Tumor Detection | Image Classification | VGG16, TensorFlow, Flask |
| Mental Health Sentiment Analysis | NLP Classification | BiLSTM, TensorFlow, Flask |
| House Price Prediction | Regression | Scikit-learn, Flask |
| Loan Approval Prediction | Classification | Scikit-learn, Flask |
End_to_End_AI_ML_projects/
├── Brain_Tumor_Image_Classification_GCP/ # Submodule → GCP production deployment
├── Mental_Health_Sentiment_Analysis_GCP/ # Submodule → GCP production deployment
├── Brain_Tumor_Detection_Image_classificaion/ # Classical ML project (original)
├── House_Price_Prediction_System/ # Classical ML project
├── loan_approval_prediction/ # Classical ML project
└── README.md
- TensorFlow, Keras, Scikit-learn
- VGG16, BiLSTM, Random Forest, XGBoost
- Transfer Learning, Two-phase Fine-tuning
- Google Cloud Platform (Cloud Run, Cloud Build, GCS, Vertex AI)
- Docker, Artifact Registry
- GitHub Actions CI/CD
- Vertex AI Model Registry
- FastAPI, Flask
- Next.js, React, TypeScript
- Tailwind CSS, HTML, CSS
GCP-deployed projects may be temporarily suspended to avoid ongoing cloud billing charges. All projects are fully functional and can be redeployed within 15–20 minutes using the CI/CD pipelines documented in their respective repositories. Production screenshots are available in each project repository as proof of working deployments.
Prem Kumar — AI/ML Engineer
GitHub: https://github.com/prem332