MLOps App Demo

A complete end-to-end project demonstrating the MLOps lifecycle, from model design to deployment and monitoring.
The goal is to build and deploy a CNN-based classifier for handwritten digits while showcasing model hosting and responsive web app integration.
Designed and trained a deep learning model for handwritten digit classification (MNIST)
Applied MLOps principles: versioning, testing, automation, monitoring, tracking and reproducibility.
Deployed the trained model to Hugging Face Spaces for easy accessibility.
Integrated MLflow locally for tracking experiments, metrics and artifacts.
Built a responsive web application to allow users to draw and classify digits in real-time.
Deployed the web app on Vercel for public access.
Documented the full workflow as a step-by-step tutorial (Tutorial Link).
Methods & Workflow:
- Data preprocessing and augmentation for robust training.
- Model design with CNN architecture.
- Training, evaluation and experiments logging using MLflow.
- Model packaging and deployment on Hugging Face.
- Frontend integration.
- Continuous improvements cycle: monitoring, retraining and redeployment.