Skip to content
Badisa Srikanth
All projects
Live demoDevOps & MLOps

Student Result Prediction (MLOps)

ML model shipped through CI, a quality gate, Docker, GHCR and Render.

Live demo (opens in a new tab) Source code (opens in a new tab)

Problem

Practise taking a machine learning model from training to a tested, containerised and deployed service.

Solution

A small prediction model with a GitHub Actions pipeline that trains, tests, gates on model quality, builds a Docker image, publishes it to GHCR and deploys it to Render.

Architecture

  1. Python training script and Flask prediction API
  2. GitHub Actions: tests, ML quality gate and artifact upload
  3. Docker image with a health check
  4. GitHub Container Registry tags: latest, v1 and commit SHA
  5. Deployed from GHCR to Render

My contribution

Built the model, tests, workflows, Dockerfile and deployment.

Implemented

  • Automated tests for the app and ML pipeline
  • ML quality gate in the CI pipeline
  • Preserved ML artifacts and a release package
  • Docker health-check failure and recovery demonstration
  • Automated GHCR publishing
  • Deployed API with a /health endpoint (free Render instance, so the first request can be slow)

Results

  • Successful GitHub Actions runs for each practical, linked in the DevOps & MLOps journey
  • Public container image on GHCR