M.Sc. in Machine Learning · MBZUAI
I study the foundations of causal inference, causal representation learning and discovery, and robust and generalizable artificial intelligence.
My current work spans causal mediation, joint counterfactual evaluation, and object-centric structured world models.
I am interested in methods that expose causal structure and remain reliable under interventions, partial information, distribution shifts, and changing environments.
I study path-specific causal effects and adaptive treatment allocation, including fixed-confidence identification of interventions that maximize natural direct outcomes.
I work on evaluating otherwise unidentified joint counterfactual functionals when only marginal distributions and edgewise correlations are available.
I develop object-centric latent representations with sparse transition dependencies for modular and generalizable models of structured dynamics.
These papers represent my current work across causal inference, counterfactual reasoning, and representation learning for long-form video understanding.
We formulate maximization of the expected natural direct potential outcome as a fixed-confidence best-arm identification problem, establish identification from interventional distributions, derive instance-dependent sample-complexity lower bounds, and develop an asymptotically optimal Track-and-Stop algorithm.
We study how partial dependence information can constrain joint counterfactual quantities when the complete joint counterfactual distribution is unavailable.
This work develops asymmetric dual projections for modular alignment between long-form video representations and text.
I am pursuing an M.Sc. in Machine Learning at MBZUAI, with research spanning causal inference, structured representations, and robust AI.
M.Sc. in Machine Learning, Abu Dhabi, UAE. Current CGPA: 3.61/4.00.
I write long-form notes in Overleaf and publish the final versions here as PDFs, with short abstracts and topic tags for quick browsing.
These research and industry experiences shaped my approach to machine learning, but the current focus of this website is my academic work in causality.
I am happy to discuss causal inference, causal representation learning, causal discovery, structured world models, and related research collaborations.