About Me

Hello! I am a PhD student in the Paul G. Allen School of Computer Science and Engineering at the University of Washington, Seattle. I am interested in exploring ways to make Machine Learning trustworthy. I have explored explainability and fairness aspects of Machine Learning to make it trustworthy. I hope to continue exploring this area in future.

I majored in Electrical Engineering from IIT Kanpur and graduated with a minor in Computer Science in 2019. At IIT Kanpur, I worked on research projects in Fairness in Machine Learning and Program Analysis of Machine Learning and C code. In my free time, I enjoy reading blogs and books, connecting with friends, and exploring outdoors.

Interests

  • Trustworthy ML
  • Explainability in ML
  • Fairness in ML

Education

  • PhD in Computer Science, 2024 (expected)

    University of Washington, Seattle

  • Masters in Computer Science, 2021

    University of Washington, Seattle

  • B.Tech in Electrical Engineering, 2019

    Indian Institute of Technology, Kanpur

Recent Posts

Amortized Counterfactual Explanations

We present a novel technique to generate counterfactual explanations by formulating the problem into an MDP and then use planning or Reinforcement Learning approaches to learn a policy. The learned policy can then generate counterfactuals for all datapoints from the training distribution using only inference, and is thus amortized. This is the first approach that satisfies all desirable desiderata of counterfactual explanations. Link to the post.
Jun 15, 2021 0 min read Explainability in ML
Amortized Counterfactual Explanations

Reinforcement Learning for Counterfactual Explanations

Within the field of explainable AI (XAI), the technique of counterfactual explainability has progressed rapidly, with many exciting developments just in the past couple years. To help crystallize and understand the major development areas, we presented a new paper at the NeurIPS Workshop on ML Retrospectives, Surveys, and Meta-Analyses. Link to the post.
Jun 15, 2021 0 min read Explainability in ML
Reinforcement Learning for Counterfactual Explanations

Publications

Quickly discover relevant content by filtering publications.

RecXplainer: Post-hoc attribute-based explanations for recommender systems

Recommender systems are ubiquitous in most of our interactions in the current digital world. Whether shopping for clothes, scrolling …
PDF Slides

Counterfactual Explanations for Machine Learning: Challenges Revisited

Counterfactual explanations (CFEs) are an emerging tech- nique under the umbrella of interpretability of machine learning (ML) models. …
PDF Slides

Pitfalls of Explainable ML: An Industry Perspective

As machine learning (ML) systems take a more prominent and central role in contributing to life-impacting decisions, ensuring their …
PDF Slides

Amortized Generation of Sequential Counterfactual Explanations for Black-box Models

Explainable machine learning (ML) has gained traction in recent years due to the increasing adoption of ML-based systems in many …
PDF Slides

Removing biased data to improve fairness and accuracy

Machine learning systems are often trained using data collected from historical decisions. If past decisions were biased, then …
PDF Slides
See all publications

Experience

 
 
 
 
 

Research Intern

Amazon

Jun 2022 – Oct 2022 San Francisco, CA

Responsibilities include:

  • Worked on interpretability of recommender systems
 
 
 
 
 

Research Fellow

Arthur AI Inc.

Jun 2020 – Sep 2021 Washington D.C., USA

Responsibilities include:

  • Reading and understanding research papers
  • Developing novel technique to solve counterfactual explainability
  • Deployment of the solution for Arthur’s clients
 
 
 
 
 

Research Intern

AST Lab, ETH Zurich

Jun 2019 – Sep 2019 Zurich, Switzerland

Responsibilities include:

  • Explored solutions to shape incompatibility bugs in Machine Learning Code
  • Tested our approach on several benchmarks
 
 
 
 
 

Research Intern

CSAIL, MIT

May 2018 – Aug 2018 Boston, USA

Responsibilities include:

  • Developing new technique which allocate bits to floating point programs
  • Comparing our technique to previous techniques
 
 
 
 
 

Research Intern

NUS School of Computing

May 2017 – Aug 2017 Singapore

Responsibilities include:

  • Analyse programs using KLEE
  • Developing tool to convert CSP programs into C code
 
 
 
 
 

Undergraduate Researcher

CSE, IIT Kanpur

May 2016 – Mar 2017 Kanpur, India

Responsibilities include:

  • Developing technique to automatically repair buggy C programs

Contact

  • vsahil@cs.washington.edu or v.sahil1@gmail.com
  • Paul G. Allen Center, 185 E Stevens Way NE, Seattle, WA 98195
  • Follow Me
  • Connect On