Picture of me

Amit Sharma

Postdoctoral Researcher
Microsoft Research, New York

Blog: A gentle introduction to causal inference
Twitter LinkedIn Google Scholar Dribbble Quora

Research Summary

My work spans a broad range of research areas, connected through a central goal of understanding human behavior through digital data. Technological advances in systems and sensors have provided an opportunity to observe what is happening (e.g., ratings, purchases) in minute detail, but it is much harder to say why. My research aims to bridge this gap between the whats and whys, employing techniques from machine learning, statistics and experimentation to discover causal patterns that lead to the activity and decisions we see.

For example, when we go shopping, how do the website's recommendations affect our final purchase? When we choose what to read, how much do our friends (or celebrities) influence our behavior? Rather than relying on untestable assumptions about the data-generating process as in traditional methods for causal inference, the fundamental idea in my research is that we can use properties of the available data to infer causal knowledge.

At Microsoft Research, I am working on building data-driven methods for causal inference and applying it to problems in both online and offline contexts. I received my Ph.D. from Cornell, advised by Dan Cosley, where I studied the role of influence and recommendation in social networks. Prior to graduate school, I studied computer science at IIT Kharagpur.

Here is my CV.


Prediction and Explanation in social systems: How complex predictive models can yield robust explanationsScience, 2017 | Paper

Data mining for Instrumental Variables: Estimating the causal impact of recommender systemsEC 2015 | Paper

Preference Model-based Matching: Distinguishing between personal preference and social influence in social activity feedsCSCW 2016 Hon'ble Mention | Paper

Simpson's Paradox in Reddit: Using Time-aware analyses to better understand behaviorWWW 2016 | Paper


April 2017

Outstanding Reviewer award at the 26th World Wide Web (WWW) conference.

March 2017

Invited speaker at the Global Data Science Conference 2017. [Interview and slides]

Feb 2017

New paper on evaluating prediction and explanation in social systems published in Science. Featured in Next@Microsoft. [PDF]

Nov 2016

Invited talk on "Causal inference in data science" at the Engineering and Data Science conference (DataEngConf), New York. [Slides]

Nov 2016

Presented work on "Auditing search engines for differential satisfaction across demographics" at the Workshop on Data Accountability and Transparency (DAT), New York University Law School. [Slides]

Oct 2016

Invited talk on "Limits of social influence in online social networks" at the Affective Brain Lab, University College London.

Jun 2016

Tutorial on causal inference at the 2nd International Computational Social Science (IC2S2) conference. Slides and code on Github!

Apr 2016

Invited talk at MIT Marketing Seminar on data mining for causal inference and its applications in estimating impact of recommender systems. [Slides]

Feb 2016

Honorable mention for Best paper award at the 19th ACM CSCW Conference for our paper on estimating social influence through social activity news feeds.

Selected Talks

Work Experience

Microsoft Research

New York, USA
July 2016 - Current

Postdoctoral Researcher in Computational Social Science.

Microsoft Research

New York, USA
May - August 2014

Worked with Jake Hofman and Duncan Watts.
Estimation of causal impact of recommender systems.

Google Inc.

Mountain View, USA
May - August 2013

Worked with Gueorgi Kossinets.
Inference of attributes for local businesses, studying the evolution of their rating.

LinkedIn Corp.

Mountain View, USA
May - August 2012

Worked with Baoshi Yan.
Novel pairwise models for learning implicit user feedback on recommendations.


Lausanne, Switzerland
May - July 2009

Worked with Frederic Kaplan and Pierre Dillenbourg.
Prosodic analysis of speech and visualizations for supporting collaborative dialogue.

IBM Research

New Delhi, India
May - July 2008

Worked with Akshat Verma and Gargi Dasgupta.
Local-optimal search algorithms for dynamic composition of web services.


Honors and Awards

Working papers


Workshop and Demos

Datasets for social recommendation and diffusion

When I started working on recommendations within social networks, there were not many datasets available that had both social connections and people's preferences. Even the ones that existed were hard to find, so I am listing out datasets that are useful for studying people's preferences in social networks and designing social recommender systems.

Contact Me