I’m a second-year master’s student in Computational Neuroscience at the University of Tübingen. I am doing my master’s thesis with Peter Dayan’s group (AGPD), under Roxana Zeraati, using meta-reinforcement learning and ecological decision-making tasks to study how recurrent agents internalise task structure in their dynamics.

Earlier this year, I did an essay rotation at the Max Planck Institute for Intelligent Systems with Bernhard Schölkopf and Lancelot Da Costa on growing causal abstractions in brains and machines. That project helped sharpen many of the questions that now drive my research.

Research

I am interested in how learning is orchestrated in the brain. In particular, I want to understand how neural dynamics, memory, and interaction with the world give rise to internal structure that supports adaptive behaviour under uncertainty.

My long-term focus is continual learning in the brain: how knowledge is built, stabilised, and reused over time without losing flexibility. I use tools from bayesian reinforcement learning and recurrent neural networks as lenses on these questions, with the broader goal of using insights from neuroscience to help guide continually learning AI.

Background

I completed my B.Tech. in Computer Engineering at Jamia Millia Islamia University, New Delhi, in 2024. My bachelor’s thesis on “Music Generation from Brain Scans”, supervised by Prof. Tanvir Ahmad, was my first major research project in neuroscience and drew me into the field. Since then, I have worked across multimodal language, self-supervised learning, visual attention, and remote sensing, with a recurring interest in learning, representation, and behaviour.

Highlights

04/2026   Began Master’s Thesis at the Computational Neuroscience Lab
04/2026   Awarded IMPRS MMFD Scholarship.
03/2026   Completed Essay rotation at MPI-IS.
12/2025   Completed Lab rotation at MPI-KYB.
10/2024   Began M.Sc. in Computational Neuroscience at Tübingen.
06/2024   Graduated with Honours in B.Tech. Computer Engineering.
03/2024   Paper on LLMs accepted at the SemEval 2024 Workshop.
08/2023   Completed Summer Research Internship at TU Munich.
07/2023   Paper on DEM super-resolution published at IEEE IGARSS 2023
01/2023   Awarded DAAD-WISE Scholarship.

Publications

Natural Building Blocks for Structured World Models: Theory, Evidence, and Scaling
L. DaCosta, S. Namjoshi, M. A. Ansari*, B. Schölkopf
World Modeling Workshop, 2026
[PDF]

JMI at SemEval 2024 Task 3: Two-step approach for multimodal ECAC using in-context learning with GPT and instruction-tuned Llama models
Arefa, M. A. Ansari*, C. Saxena, T. Ahmad
ACL SemEval Workshop, 2024
[PDF] [Code]

Master GAN: Multiple Attention is all you Need: A Multiple Attention Guided Super Resolution Network for Dems
A. Mohammed, M. Kashif, M. H. Zama, M. A. Ansari* and S. Ali
IEEE IGARSS, 2023
[PDF] [Code]

---

Theme inspired by Shruti's page