Projects · Running on K3s

Where I build
ML systems.

End-to-end machine learning projects — from raw data to production deployments. Each project is live, containerised, and running on my own Kubernetes cluster. By Amit Choubey.

🐍 Python · FastAPI · Streamlit ☸️ K3s on Hetzner VPS ☁️ AWS · MLflow · Docker
3
Live projects
4+
In progress
K3s
Self-hosted
AWS
Cloud pipelines
All Projects
Each project is independently deployed at its own subdomain.
🏠
Live
Housing Price Predictor
housing.labs.amitchoubey.dev ↗
End-to-end ML pipeline predicting house prices. Scikit-learn model served via FastAPI, interactive Streamlit UI, fully containerised on Kubernetes.
scikit-learnFastAPIStreamlitK8s
🌸
Live
Iris Classifier
iris.labs.amitchoubey.dev ↗
Classic Iris dataset explorer and classifier. Interactive Streamlit app for classification and visualisation, containerised on Kubernetes.
scikit-learnStreamlitK8s
🎬
Live
Movie Recommendation
movie.labs.amitchoubey.dev ↗
Collaborative filtering with matrix factorisation, user-item latent factors, similarity matrices, and hybrid signals. Full ML pipeline containerised with Docker, deployed on Kubernetes (Service + Ingress), served via Streamlit.
collaborative filteringmatrix factorisationDockerK8sStreamlit
🔍
Coming soon
RAG Application
Retrieval-Augmented Generation pipeline with vector search. Document Q&A with semantic chunking, reranking and context-aware LLM responses.
LangChainPineconeOpenAIFastAPI
🚨
Coming soon
Fraud Detection
Real-time transaction fraud detection on imbalanced datasets. XGBoost with SMOTE oversampling and SHAP explainability for full interpretability.
XGBoostSHAPMLflowStreamlit
💬
Coming soon
Sentiment Analyser
Fine-tuned BERT for product review sentiment classification. Deployed with ONNX runtime for optimised inference latency and a live demo UI.
BERTHuggingFaceONNXFastAPI