TECH 39 — Statistical Foundations for AI, Machine Learning, and Data Science
Quarter: Winter
Instructor(s): Liliya Lavitas
Date(s): Jan 19—Feb 23
Class Recording Available: Yes
Class Meeting Day: Tuesdays
Grade Restriction: NGR only; no credit/letter grade
Class Meeting Time: 5:30—7:00 pm (PT)
Tuition: $500
Refund Deadline: Jan 21
Unit(s): 0
Enrollment Limit: 45
Status: Registration opens Nov 30, 8:30 am (PT)
Quarter: Winter
Day: Tuesdays
Duration: 6 weeks
Time: 5:30—7:00 pm (PT)
Date(s): Jan 19—Feb 23
Unit(s): 0
Tuition: $500
Refund Deadline: Jan 21
Instructor(s): Liliya Lavitas
Grade Restriction: NGR only; no credit/letter grade
Enrollment Limit: 45
Recording Available: Yes
Status: Registration opens Nov 30, 8:30 am (PT)
AI and machine learning rely on statistical principles essential for building, evaluating, and interpreting algorithms. This course provides a rigorous grounding in the methodologies underpinning contemporary data science. Students will explore prediction and classification models, progressing from linear regression and clustering to random forests, XGBoost, PCA, and transformer architectures. We will demystify complex systems, showing how advanced large language models (LLMs) share fundamental mathematical DNA with basic linear models. The focus is on statistical scaffolding and results interpretation, not a tutorial on training LLMs or building agents.
This is a fast-paced, lecture-based course featuring live code demonstrations to show concepts in action. To maximize technical depth, there are no breakout rooms or in-class coding; datasets and notebooks are provided for independent homework.
This is a fast-paced, lecture-based course featuring live code demonstrations to show concepts in action. To maximize technical depth, there are no breakout rooms or in-class coding; datasets and notebooks are provided for independent homework.
The material involves mathematical rigor. Students should be familiar with Python, linear algebra, and basic calculus, including concepts such as gradients and maximum likelihood estimation (MLE).
LILIYA LAVITAS
Data Science Manager, Google Gemini
Liliya Lavitas leads a data science team at Google Search. She previously led data science teams for Gemini model evaluation and training. She also managed a data science team at Twitter and was an Amazon data scientist within the Alexa machine learning department, working on the internationalization of Alexa devices as well as NLU models. She received a PhD in statistics from Boston University. Textbooks for this course:
There are no required textbooks; however, some fee-based online readings may be assigned.