Sarah Tan Email & Phone Number

United States

San Francisco, California, Us

AI safety for Salesforce built generative models throughout the entire model development and deployment lifecycle.

The Women in Machine Learning (WiML) 501c3 nonprofit’s mission is to enhance the experience of women in machine learning. I have been involved with the organization since 2016, from being general chair of workshop at NeurIPS, to joining the board of directors in 2018. As VP of Events in 2019-2020 I oversaw all of WiML's events. Other efforts I led include writing WiML's Code of Conduct and launching a new funding program during the pandemic to fund underrepresented women worldwide to attend virtual conferences. Currently, as President I am focused on streamlining operations and growing the organization to serve the needs of women in different career stages and geographies.

Ithaca, Ny, Us

I have a visiting appointment in the College of Computing and Information Science to do research in medical imaging and continue my interest in healthcare.

Portland, Or, Us

Projects included building and evaluating retrieval augmented generation LLM systems for healthcare use cases and writing the company’s Responsible AI policy.

During my time at Meta I was first in the Core Data Science (now Central Applied Science) organization and then the Responsible AI organization. I developed AI bias mitigations and helped product teams apply them to their models. I also worked on harm mitigations for integrity teams, personalized recommendations, and experimentation tooling.Papers:- Error Discovery by Clustering Influence Embeddings. NeurIPS '23- Interpretable Personalized Experimentation. KDD '22- Practical Policy Optimization with Personalized Experimentation. NeurIPS '21 Workshop- Efficient Heterogeneous Treatment Effect Estimation With Multiple Experiments and Multiple Outcomes. CODE '21

San Francisco, California, Us

Collaborated with Zuckerberg hospital to investigate use of interpretable machine learning models for healthcare, and helped teach Biostats 216 (Machine Learning for the Biomedical Sciences) to clinicians and UCSF staff.

Consultant on observational causal inference methods

Redmond, Washington, Us

Two Microsoft Research internships working with Rich Caruana, Ece Kamar, and Kori Inkpen, resulting in the following papers:- Distill-and-Compare: Auditing Black-Box Models Using Transparent Model Distillation. AIES '18- Axiomatic Interpretability for Multiclass Additive Models. KDD '19- Do I Look Like a Criminal? Examining the Impact of Racial Information on Human Judgement. CHI '20- Purifying Interaction Effects with the Functional ANOVA: An Efficient Algorithm for Recovering Identifiable Additive Models. AISTATS '20- How Interpretable and Trustworthy are GAMs? KDD '21- Considerations When Learning Additive Explanations for Black-Box Models. Machine Learning Journal '23- Missing Values and Imputation in Healthcare Data: Can Interpretable Machine Learning Help? CHIL '23

At this startup research lab, I worked on various NLP problems, such as e-discovery from legal documents, modeling of movie scripts, etc. I also munged a lot of data. Some of the work won best paper awards (https://journals.sagepub.com/doi/abs/10.1177/0003122415598534).

I developed predictive models for healthcare, such as hospital readmissions, adverse drug interactions, etc. on data from all of New York City's public hospitals. Some of the work resulted in the following papers:- Hospital Readmission Rates: Related To Ed Volume, Population, And Economic Variables. Academic Emergency Medicine 2012.- Using PROC GENMOD to Investigate Drug Interactions: Beta Blockers and Beta Agonists and Their Association with Hospital Admissions. SAS Global Forum 2013.- Two Ways of Modeling Hospital Readmissions: Mixed and Marginal Models. JSM 2013.

New York, Ny, Us

Data mining on customs records to detect aberrant trade transactions. Worked with UN Head of Trade Statistics to update trade transactions reporting guidelines for UN member nations.

Doctor Of Philosophy (Ph.D.), Statistics

Bachelor’S Degree, Statistics, Economics