About Me
Dr. Dmitry Kazhdan
AI Researcher & Builder
👋About Me
Current: Senior Product Owner at Revolut, figuring out how multi-agent teams of Human and AI agents (such as LLMs / MLLMs) can reach optimal outcomes and power the next generation of financial products.
Past: Ex-Co-founder & CPTO of Tenyks (YC S21), where I helped build Visual Intelligence AI products for real-time camera analytics in the Hospitality Industry. Yes, I've seen things. I've led teams through startup hellfire, built LLM/VLM products from scratch, and lived to tell the tale. Academically, I hold a PhD from The University of Cambridge in Explainable AI, where I happily wrestled with Concept-based Explanations across multiple modalities (expertly supervised by Prof. Mateja Jamnik & Prof. Pietro Lio). Spoiler: XAI doesn't solve the Evil AI problem - turns out, you need Explainable Humans for that too.
Outside of AI: As a film enthusiast, I'm paying close attention to AI video generation efforts (e.g. Veo). I used to play classical guitar (Grade 8 Distinction) and will happily nerd out over cinematography, sci-fi plots, and existential humour.
🎓Education
PhD in Artificial Intelligence
The University of Cambridge
Enhancing Interpretability: The Role of Concept-based Explanations Across Data Types
MEng in Computer Science
The University of Cambridge
MARLeME: A Multi-Agent Reinforcement Learning Model Extraction Library
High-School
The Heritage Private School, Cyprus
A-levels: 4A*, 1A; AS-levels: 1A; IGCSEs: 9A*, 1A
💼Work Experience
Senior Product Owner (Data Science)
Revolut
2025 - Present
Applying latest advancements in LLMs / Agentic AI to Financial Crime detection & prevention
Co-Founder and CTO
Tenyks
2021 - 2025
Lead R&D of a cutting-edge camera stream analytics platform for the Hospitality Industry
Research Intern
Nokia Bell Labs
2019
Explored Variational Continual Learning for personalised ML systems using sparse user data
📚Recent Publications
Now You See Me (CME): Concept-based Model Extraction
Conference on Information and Knowledge Management (CIKM), AIMLAI Workshop
MEME: Generating RNN Model Explanations via Model Extraction
Advances in Neural Information Processing Systems (NeurIPS), HAMLETS Workshop
MARLeME: A Multi-Agent Reinforcement Learning Model Extraction Library
International Joint Conference on Neural Networks (IJCNN)