Simon Ejdemyr – Data Scientist
Things I do |
Writing |
Tutorials
I'm a data scientist and (self-proclaimed) engineer building
tools for decision making and strategy. Since 2020, I've been at
Netflix focusing on experimentation and observational causal
inference.
My background is in computational social science. Although I use
a range of quantitative methods in my work, I believe the best
insights and predictions are formed when large-scale data and
automation are combined with careful human judgement and
curation.
I'm from Sweden 🇸🇪, where I enjoyed a short pro football
(fine... soccer) career before going to the US for college and a
PhD. I live in Los Angeles with my wife, son, and daughter. My
last name is pronounced AY-duh-meer. Feel free to connect on
LinkedIn.
Things I do
- Build products and tools to make sense of complex data
- Design systems for randomized controlled trials and
observational causal inference
- Develop, prototype, and productize machine learning and
inference models based on analytical solvers or (Bayesian)
probabilistic programming
- Use Python, R, SQL, bash, Stan, and more; and contribute to
and design APIs
- Enjoy teaching and mentoring (in 2016 I won Stanford's Centennial Teaching
Award)
Writing
A few things I've published in academia
and industry:
-
(2024)
Learning the Covariance of Treatment Effects Across Many
Weak Experiments. Proceedings of the 30th ACM SIGKDD
Conference on Knowledge Discovery and Data Mining, with
Aurélien Bibaut, Winston Chou, and Nathan Kallus. [Blog
Post]
-
(2023) Long-Term
Causal Inference with Imperfect Surrogates using Many Weak
Experiments, Proxies, and Cross-Fold Moments. MIT CODE,
with Aurélien Bibaut, Nathan Kallus, and Michael Zhao.
-
(2022) A
Framework for Generalization and Transportation of Causal
Estimates Under Covariate Shift. MIT CODE, with Apoorva
Lal and Wenjing Zheng.
-
(2021)
Decision Making at Netflix. Netflix Tech Blog, with
Martin Tingley and others.
-
(2020)
Low-latency Multivariate Bayesian Shrinkage in Online
Experiments. MIT CODE, with Matthew Wardrop and Martin
Tingley.
-
(2019)
4 Keys to Using Machine Learning for Campaign Measurement.
Facebook IQ blog.
-
(2018)
Do Elections Improve Constituency Responsiveness? Evidence from U.S. Cities.
Political Science Research and Methods, with Darin Christensen.
-
(2017)
Segregation, Ethnic Favoritism, and the Strategic Targeting
of Local Public Goods. Comparative Political Studies,
with Eric Kramon and Amanda Lea Robinson.
-
(2015)
Global, Regional, and National Levels and Trends in Under-5
Mortality Between 1990 and 2015, with Scenario-based
Projection to 2030. The Lancet, with UNICEF colleagues.
Tutorials
At Stanford I taught classes in applied statistics, for which I
developed a number of R tutorials. While likely out of date,
I've heard some are still useful.