ta-training

Technical Handbook

This section documents the technical standards, workflows, and review practices that TAs and mentors are expected to enforce in the Data Science Clinic.

Purpose

• Provide TAs and mentors with a reference for evaluating and reinforcing best practices.
• Support consistency in repo hygiene, code review, and reproducibility across projects.
• Equip TAs to give actionable, constructive feedback on technical issues.

Contents

• Python Best Practices
  Common Python pitfalls and language-specific patterns TAs should watch for: mutable defaults, scoping issues, error handling.

• Code Style and Structure
  Guidelines for organizing code, naming conventions, documentation standards, and professional presentation.

• Data Validation and Types
  Type annotations and Pydantic validation patterns for ensuring data quality and code clarity.

• Configuration Management
  Managing settings, secrets, and environment-specific configuration with Pydantic Settings.

• Debugging and Testing
  Disciplined debugging practices and lightweight testing approaches for data science projects.

• Code Review
  Guidelines for TAs when reviewing pull requests and student repos.

• AI Usage Guidelines
  Guidelines for handling AI-assisted code in student projects and code reviews.

• GitHub Repo Management
  Branching strategies, PR workflows, and common scenarios TAs encounter with student repositories.

• Docker and Make
  Container setup patterns and Make targets for reproducible data science workflows.

• Dependencies and Environments
  Managing Python dependencies, Docker environments, and reproducible setups.

• Paths and I/O in Containers
  Path handling in Docker environments and separating I/O operations from computation logic.

• Interview Exercise
  Code review prompt used in TA interviews and training.

• Examples
  Small runnable setups for TAs to practice before working with students.

How to Use This Section

• Use these documents when preparing to review student repos or pull requests.
• Refer back when you need talking points for TA sessions (e.g., how to coach Docker, paths, or PR reviews).
• In final repo reviews, use these standards to grade consistently with the official repo evaluation checklist.

Key Principles

• Enforce, don’t code: TAs guide students to solve issues, not fix repos themselves.
• Consistency: Apply the same standards across projects so grading is fair.
• Professionalism: Model how real data science teams work.
• Escalation: Raise systemic or recurring issues with mentors and the Clinic Director.