Summary
Daniil Cherniavskii is a PhD student at the QUVA Lab, University of Amsterdam, with a decade of experience applying machine learning to NLP, embodied AI, and materials science. He focuses on improving reasoning, planning, and code-generation in large language models, bridging theoretical advances with real-world adaptability. His prior roles include research at AIRI—where he used topological data analysis to probe attention maps, intrinsic dimensionality of synthetic text, and topology-preserving dimensionality reduction—and internships at Huawei and Meta exploring adversarial text generation and discrete diffusion models. Daniil combines a strong physics and applied mathematics foundation from MIPT with an MS in Data Science, enabling rigorous, cross-disciplinary approaches to robustness and interpretability. He is particularly interested in using geometric and topological insights to reveal model failure modes and improve generalization in embodied and language-driven tasks.
10 years of coding experience
2 years of employment as a software developer
Master of Science - MS, Data Science, Master of Science - MS, Data Science at Skolkovo Institute of Science and Technology
Doctor of Philosophy - PhD, Deep Learning, Doctor of Philosophy - PhD, Deep Learning at University of Amsterdam
Bachelor of Applied Science - BASc, Physics and Applied Mathematics, Bachelor of Applied Science - BASc, Physics and Applied Mathematics at Moscow Institute of Physics and Technology (State University) (MIPT)