DS/CS‑549: Spark! Machine Learning Practicum

🎓 • Boston University

Course Overview

Applied practicum where student teams build machine learning models for real-world clients. Emphasizes the full ML lifecycle—from problem scoping and dataset curation to model development, evaluation, and stakeholder presentation.

Teaching Impact

September 2018 – Present: Taught early offerings and later transitioned to an oversight role. Recruited and supported new instructors starting in Spring 2023. Regular guest lecturer, particularly on project scoping and applied ethics.

Prepares students for careers in applied ML by embedding them in industry- or research-driven team projects. Closely coordinated with Spark and parallel to the DS/CS‑519 software practicum structure.

Learning Objectives

Students develop end-to-end machine learning skills through:

  1. Problem Scoping: Translating business problems into ML problems
  2. Data Pipeline Development: Collection, cleaning, and feature engineering
  3. Model Development: Selection, training, and validation of ML models
  4. Evaluation and Interpretation: Performance metrics and model explainability
  5. Ethical ML: Bias detection, fairness, and responsible deployment
  6. Stakeholder Communication: Presenting technical results to non-technical audiences

Course Structure

  • Real-world Projects: ML solutions for actual business or research challenges
  • Team Collaboration: 3-4 student teams with diverse skill sets
  • Full ML Lifecycle: From data exploration to model deployment
  • Client Engagement: Regular meetings with project stakeholders
  • Applied Ethics: Emphasis on responsible ML practices throughout

Key Focus Areas

  • Applied Machine Learning: Practical implementation over theoretical depth
  • Project Scoping: Learning to identify appropriate ML applications
  • Data Ethics: Addressing bias, fairness, and transparency
  • Professional Communication: Translating technical work for business audiences
  • Industry Tools: Experience with production ML frameworks and tools

Spark Program Integration

Parallel structure to the software engineering practicum, providing students with comprehensive experience in both traditional software development and modern ML applications.