TechQA Agent — Agentic RAG
Agentic RAG that answers IBM technical-support questions across 28k Technotes. When single-shot retrieval misses, the agent rewrites the query and searches again before answering — grounded in retrieved docs and measured against gold answers.
- Agentic RAG
- Qdrant
- MiniLM
- LLM
- Python
View on GitHub →Campaign Success Prediction (MLOps)
End-to-end MLOps pipeline predicting marketing-campaign success. A Random Forest classifier tuned and versioned in MLflow, wrapped with experiment tracking, orchestration, deployment, and monitoring for a reproducible, production-style workflow.
- MLOps
- MLflow
- Random Forest
- Python
- Pipeline
View on GitHub →Steel Industry Energy Classification
Classification model predicting energy-load type for a smart small-scale steel plant (UCI dataset), letting the plant anticipate high/medium/low consumption and cut costs. Covers EDA, model training and tuning, dependency management, and containerization.
- Machine Learning
- Classification
- Python
- Docker
View on GitHub →