Milestone 5: Final Project Delivery

These guidelines provide direction for preparing Milestone 5. As always, every project is unique — if your work doesn’t align perfectly with these expectations, discuss it with your TF.

Milestone 5 focuses on bringing your project from a locally functional system (Milestone 4) to a production-grade deployed application, along with preparing your public-facing communication materials. By the end of this milestone, your system should be fully deployed on cloud infrastructure, demonstrate scalability, and be ready to present during the course showcase.

This milestone also emphasizes clear communication: a video walkthrough, a public-facing blog post, and preparation for your booth during the live event.

Key dates:

  • Showcase: 12/10
  • Due Date: 12/11

Objectives

  1. Production Deployment & Infrastructure Automation

    • Deploy your full application to a cloud Kubernetes cluster (GCP or AWS).
    • Demonstrate reliability and basic scalability (e.g., scaling pods up/down under load).
    • Automate infrastructure provisioning and deployment using Pulumi.
    • Integrate ML workflow elements (data preprocessing, training, evaluation, retraining triggers) into your deployed system.

  2. CI/CD for Production

    • Extend your GitHub Actions pipelines to support deployment.
    • Include unit tests and integration tests tests.
    • Achieve minimum 60% test coverage, and clearly document what remains untested.
    • Ensure merges to main trigger an automated build-and-deploy pipeline.

  3. Public Communication and Presentation Materials

    • Prepare a polished 6-minute video describing and demonstrating your project.
    • Write a 600–800 word Medium blog post explaining your project to a general audience.
    • Create engaging booth materials for the showcase (e.g., diagrams, QR codes).

  4. Showcase Preparation

    • Ensure your application is publicly accessible, stable, and easy to demo.
    • Prepare to explain your architecture, decisions, and business value to visitors.

Deliverables

1. Technical Implementation

  • Kubernetes Deployment:
    • Deploy the application to a Kubernetes cluster.
    • Demonstrate basic scaling behavior by varying the load and showing how the cluster responds (e.g., scaling replicas/pods).
  • Pulumi Infrastructure Code:
    • Use Pulumi to automate the provisioning and deployment of your infrastructure (e.g., Kubernetes cluster, networking, storage, configurations, etc.) and application.
  • CI/CD Pipeline Implementation (GitHub Actions):
    • Set up a CI/CD pipeline using GitHub Actions. The pipeline should:
      • Have a unit test suite for each service/container.
      • Run integration tests on the code base .
      • Deploy updates to the Kubernetes cluster upon merging changes into the main branch.
      • Achieve at least 60% line coverage. Document which functions and modules are not covered by tests.
  • Machine Learning Workflow:
    • Demonstrate a production-ready ML workflow, including:
      • Data preprocessing, model training, and evaluation steps integrated into the pipeline.
      • Automated retraining and deployment triggered by new data or updates to the codebase.
      • Validation checks to ensure only models meeting performance thresholds are deployed.

2. Documentation

  • GitHub Repository:
    • Include a well-structured and modular codebase.
    • Provide a comprehensive README file with the following sections:
      • Prerequisites and setup instructions.
      • Deployment instructions.
      • Usage details and examples.
      • Known issues and limitations.
    • Submit the main branch for this deliverable.

3. Presentation Materials

  • Video Presentation:
    • Record a 6-minute video covering:
      • Problem statement and the proposed solution.
      • Technical architecture and key components.
      • Live demo of the application in action.
      • Challenges faced and solutions implemented.
    • Submit the video in MP4 format or YouTube link with a minimum resolution of 720p.
  • Blog Post:
    • Write a 600–800 word Medium blog post summarizing your project for a general audience. The post should highlight the problem, solution, technical approach, and impact.
    • Include visuals or diagrams where appropriate.
  • Self and Peer Review Forms:
    • Complete self-assessment and peer evaluation forms to provide feedback on team contributions.

4. Showcase (Dec 10th)

  • Event Format:
    • Each team will have 45 minutes during the live showcase to present their project.
    • Participants will visit your booth to interact with your application and learn about your implementation.
    • Monitors will be provided to most teams. Additional equipment or materials must be arranged by the team.
  • App Requirements:
    • The app must be fully functional and hosted on GCP or AWS, accessible via a public URL.
    • Include a QR code linking to your application so visitors can easily access and explore it.
    • Be prepared to explain your problem, solution, technical implementation, and business value to participants.
  • Best of Show Award:
    • A committee will evaluate all projects during the showcase to select the Best of Show. Evaluation criteria include:
      • Innovation and impact.
      • Technical complexity and robustness.
      • Clarity of presentation and engagement with participants.

Submission Instructions

  • Submit all deliverables (GitHub repository link, video file, blog post link, and self/peer review forms) via the course submission portal by 09:00 PM, December 11th.
  • No late submissions.

Evaluation Criteria

  1. GitHub (30%): Technical depth, content, clarity, coding style.
  2. Documentation (20%): README and technical report are clear, complete, and easy to follow.
  3. Presentation and Showcase (35%):
    • Video and blog post effectively communicate the project’s value and technical details.
    • Engagement during the live showcase demonstrates clarity and understanding.
  4. Self and Peer Review (15%) Fair contribution across team members.