Course Staff

Instructor: Milind Tambe

Office: Calendly Link

Office hours: 3:30-4:30 Friday

Email: [email protected]

Head TF: Kevin Huang

Office hours:

Email: [email protected]

TF: Lauren Cooke

Office hours:

Email: [email protected]

TF: Kidist Alemu

Office hours:

Email: [email protected],edu

📜 Course Description

Machine learning promises a bright future through solving previously unsolvable problems. Yet, the gap between a theoretical understanding of machine learning concepts and a practical application of these tools can be fraught with technical and ethical problems. In this course, teams of 3-5 students will work with real partner organizations to use machine learning techniques on a directly impactful project. Students will learn how to effectively explore data, create and iterate on real models, communicate and work with external partners, and incorporate ethics into their technical work. In addition, the class will include guest lectures from experts in various fields of the social impact tech space.

📄 Spring 2023 Projects

• WattTime (an environmental nonprofit that is part of the ClimateTRACE coalition): Building a model to identify and count buildings on industrial-sized livestock operations using satellite data to help model the number of animals and emissions.
• Woebot Health (a nonprofit developing a free, therapeutic, conversational agent): Building a model to classify messages as “crisis” or “non-crisis” language to better provide support resources during moments of crisis.
• Khushi Baby (a public health nonprofit supporting follow-up to primary health care in last-mile settings in India): Building an extension to Khushi Baby's dashboard to predict the probability of public health outcomes using spatial predictors/covariates.

🗝 Enrollment

Prerequisite(s): CS181/182 or equivalent and/or CS109b. Note that enrollment in this course is by application only. Students must apply through this form by 5pm, January 16.

🏆 Grading

Breakdown

Attendance & Participation: 10%

Course Assignments: 40%

• Engineering ledger: 20%
• Write-ups + Blog post (data exploration, model proposal, final): 15%
• Impact analysis: 5%

Partner Organization Project: 50%

• Midpoint presentation: 10%