This repository is a playground to learn, test, and understand MLOps concepts using the MLflow library. It is intended for hands-on experiments that span the full machine learning lifecycle — from local experimentation and tracking to model packaging and production-ready deployment.
Beyond MLflow and classic MLOps practices, this repo will also explore related topics including:
- AI agent observability and tracing (techniques for monitoring and understanding agent behavior)
- Experiment tracking, lineage, and reproducibility
- Model registry and deployment patterns
- LLMOps (operational practices specific to large language models)
- Integrations (examples with Databricks, Hugging Face, and other platforms)
- Instrumentation, monitoring, and performance profiling for ML systems
What you'll find here
- Jupyter notebooks and examples that demonstrate MLflow tracking and model management
- Tests and demos for observability, tracing, and agent-level diagnostics
- Notes and experiments related to LLMOps and agentic AI
Getting started
-
Clone the repository:
git clone https://github.com/kailas711/mlops-with-mlflow.git
-
Create a virtual environment and install dependencies (examples):
python -m venv .venv source .venv/bin/activate pip install -r requirements.txt
-
Open notebooks in the
notebooks/directory to run experiments locally.
Contributing
This repository is a personal learning playground. Contributions, suggestions, or issues are welcome — especially if you have experiments, demos, or notes on MLflow, observability, or LLMOps that could help make the repository more useful.
Goals
- Build reproducible examples using MLflow
- Capture practical notes on observability and tracing for agentic AI
- Explore operational best practices for LLMs and production ML systems