I am a Ph.D. candidate in Electrical and Computer Engineering at Purdue University, advised by Prof. Murat Kocaoglu and Prof. Mahsa Ghasemi. My research lies at the intersection of causal inference, sequential decision making, reinforcement learning, and machine learning theory. I am particularly interested in developing principled methods that can learn efficiently from limited interventional or sequential data while providing theoretical guarantees.
Open to postdoc, faculty-track, and industry research roles
Muhammad Qasim Elahi
Ph.D. Candidate in Electrical and Computer Engineering
I develop algorithms and theory for causal inference, causal discovery, causal bandits, and reinforcement learning, with a focus on information-efficient decision making under uncertainty. My long-term research goal is to build reliable decision-making systems that reason causally, learn from limited data, and provide theoretical guarantees in high-stakes settings.
Current role
Ph.D. Candidate, Purdue ECE
Research focus
Causal inference, bandits, RL, ML theory
Selected recognition
ICML Spotlight · NeurIPS publications
Opportunities
Postdoc, faculty-track, and research scientist roles
About
About me
Education
Ph.D. in Electrical and Computer Engineering
Purdue University · 2022 – 2026 expected
Advisors: Prof. Murat Kocaoglu and Prof. Mahsa Ghasemi · CGPA: 4.00/4.00
M.S. in Electrical Engineering
American University of Sharjah · 2020 – 2022
Thesis: MRAC-Based Electric Vehicle Energy Management · CGPA: 4.00/4.00
B.S. in Electrical Engineering
University of Engineering and Technology, Lahore · 2015 – 2019
Senior design: SIMO smart antenna system · CGPA: 3.91/4.00 · Rank: 5/160
Research
Research interests
Causal Bandits and Causal Reinforcement Learning
Learning decision-making policies under interventions, causal structure, uncertainty, and limited feedback.
Causal Discovery and Online Experimental Design
Adaptive intervention design and sample-efficient algorithms for learning causal graphs and causal relationships.
Information-Efficient Decision Making
Bayesian learning, regret analysis, Thompson sampling, information-directed sampling, and efficient exploration.
Human Feedback and Multi-objective Reinforcement Learning
Reinforcement learning methods that incorporate richer feedback signals, multiple objectives, and reliable policy selection.
News
Recent updates
- 2026: Awarded the Bilsland Dissertation Fellowship by the Purdue University Graduate School.
- 2026: Recognized as an ICML Gold Reviewer.
- 2025: Continued research on causal discovery, causal inference, and reinforcement learning from human feedback.
- 2024: Presented an ICML Spotlight paper on adaptive online experimental design for causal discovery.
- 2024: Two papers accepted at NeurIPS on causal bandits and Bayesian learning of causal graphs.
Publications
Selected publications
TMLR
2025
Causal Inference
First Author
Identification of Average Outcome under Interventions in Confounded Additive Noise Models
Studies when interventional average outcomes can be identified in confounded additive-noise models, clarifying what causal quantities can be recovered under structural assumptions.
NeurIPS
2025
Causal Discovery
Equal Contribution
Characterization and Learning of Causal Graphs from Hard Interventions
Provides characterizations and learning methods for recovering causal graph structure from hard interventions, with an emphasis on sample-efficient causal discovery.
L4DC
2025
Reinforcement Learning
Equal Contribution
Reinforcement Learning from Multi-level and Episodic Human Feedback
Develops reinforcement learning methods that can use richer human feedback signals across multiple levels and episodes, extending standard preference- or reward-based feedback models.
NeurIPS
2024
Causal Bandits
First Author
Partial Structure Discovery is Sufficient for No-regret Learning in Causal Bandits
Establishes that full causal graph recovery is not always necessary for no-regret learning in causal bandits; partial structure discovery can be sufficient for efficient decision making.
NeurIPS
2024
Bayesian Learning
Equal Contribution
Sample Efficient Bayesian Learning of Causal Graphs from Interventions
Introduces Bayesian methods for efficiently learning causal graphs from interventional data, targeting sample-efficient structure learning under intervention budgets.
ICML Spotlight
2024
Spotlight
Top 3.5%
Equal Contribution
Adaptive Online Experimental Design for Causal Discovery
Designs adaptive online intervention-selection methods for causal discovery, showing how limited experimental budgets can be used more efficiently.
NeurIPS
2023
Structural Causal Bandits
Approximate Allocation Matching for Structural Causal Bandits with Unobserved Confounders
Studies approximate allocation matching for structural causal bandits under unobserved confounding, addressing decision making when causal structure includes hidden common causes.
Research Themes
Selected research themes
No-regret Learning in Causal Bandits
Algorithms and theoretical analyses for causal bandit settings where partial structure discovery can enable efficient learning.
Adaptive Experimental Design for Causal Discovery
Sequential intervention-selection methods for discovering causal structure efficiently under experimental constraints.
Learning from Human Feedback
Reinforcement learning frameworks that use multi-level and episodic human feedback for improved policy learning.
Electric Vehicle Energy Management
Earlier research on adaptive control, modeling, and optimization for Li-ion batteries and electric vehicle traction systems.
Awards
Awards and recognitions
Bilsland Dissertation Fellowship, Purdue University Graduate School, 2026–2027
Gold Reviewer, ICML 2026: recognized among the top 25% of reviewers
Student Travel Grant, NeurIPS 2025
Student Travel Grant, NeurIPS 2023
Academic Excellence Award, College of Engineering Annual Awards 2023, American University of Sharjah
Dr. P. Carter Speers Medal for First Position in Intermediate Pre-Engineering, Forman Christian College, Lahore
Experience
Research and teaching experience
Graduate Research/Teaching Assistant
School of Electrical and Computer Engineering, Purdue University
Indiana, USA
- Conduct research on causal inference, causal discovery, causal bandits, and causal reinforcement learning.
- Develop algorithms, theoretical analyses, and Python-based experimental pipelines for bandit and reinforcement learning methods.
- Teaching support for Python for Data Science and Reinforcement Learning: Theory & Algorithms.
Graduate Research/Teaching Assistant
College of Engineering, American University of Sharjah
Sharjah, UAE
- Worked on control and modeling of Li-ion batteries and electric vehicle traction systems.
- Assisted students with laboratory experiments, course projects, and grading in electrical engineering courses.
Research Service
Reviewer service
I have served as a reviewer for major machine learning, artificial intelligence, statistics, and control venues.
CV
Curriculum vitae
Full academic CV
Download my complete CV for publication details, education, teaching experience, professional service, awards, skills, and references.
Contact
Contact and links
I am open to research conversations and future opportunities.
Please reach out for postdoctoral, faculty-track, industry research, or collaboration opportunities related to causal inference, reinforcement learning, bandits, and machine learning theory.