About me

Hi! I am a Senior Researcher at Qualcomm AI Research (Amsterdam). I recently completed my PhD in the Computational Intelligence group at Vrije Universiteit Amsterdam under the supervision of Jakub Tomczak and Max Welling.

My research focuses on LLM Efficiency, specifically developing efficient architectures with a focus on latent reasoning, compressed KV-cache, and sub-quadratic attention**. I am also interested in Deep Generative Modeling and Geometric Deep Learning for scientific domains.

Previously, I held research positions and internships at Microsoft AI4Science in Amsterdam and Cambridge, Qualcomm AI Research, and NVIDIA.

In my free time, I enjoy ballet and jazz dancing, playing the piano, and running.

Latest News

• 2025 I joined Qualcomm AI Research as a Senior Researcher focusing on LLM Efficiency.
• September 2025 Our paper “KaVa: Latent Reasoning via Compressed KV-Cache Distillation” was accepted to ICLR 2026.
• 2025 I successfully defended my PhD thesis: “Mapping the Invisible: Insights into Deep Latent Variable Generative Models”.
• May 2024 New paper “Hierarchical VAE with a Diffusion-based VampPrior” accepted to TMLR.
• February 2024 I started a part-time Associate Researcher position with the Machine Learning Force Field team at Microsoft AI4Science.