Applied Sciences Award
Ondrej Kobza
Czech Technical University in Prague
Ondrej Kobza is a researcher at Czech Technical University in Prague, Czech Republic, whose published work addresses conversational artificial intelligence, generative language models, secure coding assistants, and dialogue systems. His research connects natural language processing with model efficiency, safety, evaluation, and practical real-world conversational applications across evolving artificial intelligence systems. [1] [2] [3]
Abstract
Ondrej Kobza is a researcher at Czech Technical University in Prague whose work focuses on natural language processing, conversational artificial intelligence, generative language models, and secure coding assistants. His publications examine dialogue management, socialbot conversations, generative model integration, conversational enhancement, and security code generation. Research introduces AlquistCoder, a coding assistant trained with synthetic data and alignment methods, and benchmarks for evaluating secure and responsible code generation. Earlier studies address Alquist 5.0 and improvements to BlenderBot 3, emphasizing dialogue quality, model efficiency, system architecture, and evaluation. These publications collectively demonstrate an applied research direction connecting language technologies with artificial intelligence systems. [1] [2] [3]
Keywords
Natural Language Processing; Artificial Intelligence; Generative AI; Conversational AI; SocialBots; Secure Coding Assistants; Large Language Models; Dialogue Systems; Synthetic Data; Model Evaluation.
Introduction
Ondrej Kobza is a researcher at Czech Technical University in Prague, Czech Republic, whose published work addresses conversational artificial intelligence, generative language models, secure coding assistants, and dialogue systems. His research connects natural language processing with model efficiency, safety, evaluation, and practical real-world conversational applications across evolving artificial intelligence systems. [1] [2] [3]
Research Profile
Kobza’s research profile centers on natural language processing and applied generative AI, with publications spanning conversational agents, language-model enhancement, and security-oriented code generation. His work includes collaborations within the Czech Technical University research environment and examines methods for improving model behavior, efficiency, evaluation, and robustness across modern language technologies today. [1] [2] [3]
Research Contributions
Kobza has contributed to research on dialogue management, generative conversational systems, and secure coding assistants. His publications describe approaches involving dialogue trees, generative models, synthetic training data, alignment techniques, benchmark development, and system optimization methods. Collectively, these contributions address both capability and responsible deployment considerations within modern language-model research today. [1] [2] [3]
Publications
Kobza’s publication record includes studies on AlquistCoder, Alquist 5.0, and enhancements to BlenderBot 3. These works address secure code generation, socialbot conversations, conversational model architecture, evaluation, and performance optimization. The publications demonstrate research interest in applying language technologies to practical systems while investigating methods for improving reliability, efficiency, and safety. [1] [2] [3]
Research Impact
The documented research provides contributions to natural language processing through publicly described methods, evaluations, and model-development practices. The AlquistCoder study introduces synthetic-data and benchmark resources for secure coding assistants, while earlier work examines conversational architectures and model improvements. Together, these studies provide technical directions for further research in language-model systems. [1] [2] [3]
Award Suitability
The documented publication record aligns with an Applied Sciences Award focused on applications within natural language processing and artificial intelligence. Kobza’s work combines methodological development with practical system evaluation, covering conversational agents and secure code generation. The evidence supports consideration of documented work through its technical scope, applied orientation, and contributions. [1] [2] [3]
Conclusion
Ondrej Kobza’s documented research reflects an interdisciplinary application of natural language processing to conversational systems, generative models, and secure coding. His publications demonstrate engagement with model development, evaluation, and applied artificial intelligence research. The record provides a basis for recognizing contributions that connect language technology research with practical computational applications. [1] [2] [3]
External Links
References
- Kobza, O., Černý, A., Dostál, I., Šedivý, J., Rigaki, M., Sladić, M., & Garcia, S. (2026). AlquistCoder: A synthetic data approach to training compact secure coding assistants and building security benchmarks. Computational Intelligence, 42(4), e70282.
https://doi.org/10.1111/coin.70282
- Kobza, O., Herel, D., Cuhel, J., Gargiani, T., Marek, P., & Sedivy, J. (2024). Alquist 5.0: Dialogue trees meet generative models, a novel approach for enhancing SocialBot conversations. Future Internet, 16(9), 344.
https://doi.org/10.3390/fi16090344
- Kobza, O., Herel, D., Cuhel, J., Gargiani, T., Pichl, J., Marek, P., Konrad, J., & Sedivy, J. (2023). Enhancements in BlenderBot 3: Expanding beyond a singular model governance and boosting generational performance. Future Internet, 15(12), 384.
https://doi.org/10.3390/fi15120384
