I’m a Postdoctoral Researcher at Erasmus University Rotterdam, specializing in computational social science. I apply artificial intelligence and machine learning to address critical challenges in politics and governance, with a focus on democratic backsliding, digital censorship, and political conflict.
- 🔭 I'm currently working on the Horizon Europe-funded TWIN4DEM project, where I build machine learning frameworks and 'digital twins' to model and simulate threats to democratic institutions.
- 🌱 I'm currently implementing advanced AI architectures (including transformer models and reinforcement learning) to detect signals of executive aggrandizement in political text.
- 💬 Ask me about computational social science, large language models (LLMs) for political analysis, and digital governance.
- 📫 How to reach me: meher@essb.eur.nl
- Python
- R
- Julia
- SQL, STATA, LaTeX, Git
- Machine Learning (NLP, Reinforcement Learning)
- Computational Social Science
- Democratic Backsliding & Digital Governance
- Political Conflict Analysis
- International Political Economy
- Developing 'digital twins' of political systems to model and preempt threats of democratic backsliding.
- Fine-tuning Large Language Models to detect nuanced political phenomena like executive aggrandizement.
- Analyzing internet censorship and control strategies across different regime types.
- My foundational work includes co-developing ConfliBERT and ConflLlama, specialized language models for political conflict analysis.
- TWIN4DEM Project: My current research focus on strengthening democratic resilience in Europe.
- ConfliBERT & ConflLlama: Foundational language models for political science.
- Event Data Annotation with RL: A framework for structured data annotation using reinforcement learning.
- Research Papers: Including work on internet censorship, investment treaties, and political communication.
