Capstone
Cloud & DevOps
Cloud-Native CI/CD Platform
GitOps pipeline with Kubernetes, Helm and progressive delivery
- Kubernetes
- Docker
- Helm
- Argo CD
- +2
₹5,299₹8,999
4.8/74 reviews
Cloud & DevOps · Capstone
MLOps Pipeline with MLflow & DVC
Versioned data, tracked experiments and a model that retrains itself
5-6 weeks build
Overview
The machinery around a model rather than the model itself. DVC versions the dataset and defines a reproducible pipeline from raw data to trained model, MLflow records every run's parameters and metrics, and the registry promotes a model to production only when it beats the current one. A scheduled job checks incoming data for drift and retrains when it appears. Everything runs locally or on free CI minutes.
What makes it stand up
- 01Dataset versions tracked by DVC alongside the code that used them
- 02Reproducible pipeline stages rerun only when their inputs change
- 03Every experiment's parameters, metrics and artefacts in MLflow
- 04Registry promotion gated on beating the current production model
- 05Drift checks that trigger retraining, with the decision logged
Module breakdown
- 01
Data versioning
DVC tracking with a local or MinIO remote.
- 02
Pipeline
dvc.yaml stages for preparation, training and evaluation.
- 03
Tracking
MLflow runs compared side by side in its UI.
- 04
Registry and serving
Staged model versions served by a FastAPI endpoint.
- 05
Retraining
Scheduled drift check and retrain workflow in GitHub Actions.
After this, you will be able to
- Explain why code versioning alone cannot reproduce a model
- Defend your promotion rule and drift threshold
- Demo drifted data triggering a retrain and a new registered model
Technology stack
- Python
- MLflow
- DVC
- scikit-learn
- FastAPI
- GitHub Actions
- Level
- Capstone
- Domain
- Cloud & DevOps
- Build time
- 5-6 weeks
- Difficulty
You receive
- Complete, commented source code
- 60-90 page project report (IEEE format)
- Editable presentation deck
- Architecture, ER and UML diagrams
- Local setup and deployment guide
- Sample dataset or seed data
- Viva question bank with answers
- One 45-minute walkthrough call
- 30 days of doubt-clearing support