The InfoQ Podcast

Wenjie Zi on Technology and Organizational Aspects for ML Project Success

11 snips
Apr 16, 2025
Wenjie Zi, a Senior Machine Learning Engineer at Grammarly and co-founder of the Toronto AI practitioners network, shares her insights on the hurdles many ML projects face. She discusses the critical gaps in communication between business teams and ML practitioners and how to bridge them. Wenjie highlights common pitfalls like data quality issues and the importance of setting realistic expectations. She also dives into the influence of emerging generative AI systems, stressing the value of community engagement in the rapidly evolving AI landscape.
Ask episode
AI Snips
Chapters
Books
Transcript
Episode notes
ADVICE

Choose the Right Problems

  • Start by tackling the right problem, ensuring an ML solution is actually necessary.
  • A simpler, rule-based solution might suffice, or the company may not be AI-ready.
INSIGHT

Data Quality is Crucial

  • In ML, 'garbage in, garbage out' highlights the criticality of data quality.
  • Inconsistent formats, mislabels, and biased sampling can silently cause project failure.
ANECDOTE

Offline vs. Online Success

  • Offline success doesn't guarantee online success, as seen in an image recommendation project.
  • Data sampling and differing evaluation metrics between offline and online caused misalignment.
Get the Snipd Podcast app to discover more snips from this episode
Get the app