Debrecen, Hajdú-Bihar, Hungary
I build production-grade AI systems. My focus is multi-agent architectures, LLM integration, and full-stack AI backends that solve real problems. Recent work: MoleScan, a clinical AI backend for skin lesion classification using a fine-tuned Vision Transformer with LayerNorm test-time adaptation for deployment robustness. Built with FastAPI, Dockerized, and supervised by Prof. Balazs Harangi at the University of Debrecen. Target venue: CITDS 2026. PS3C Robust Inference, a three-stage framework extending a medical AI challenge with architecture-aware test-time adaptation across a heterogeneous seven-model ensemble, sample-adaptive ensembling, and conformal selective prediction with formal coverage guarantees. JEPA-RobustViT, my BSc thesis on predictive self-supervised Vision Transformers under distribution shift. Co-supervised by the University of Debrecen and BMW Q-Lab Debrecen. Beyond engineering I designed and teach the University of Debrecen's first Introduction to Quantum Computing course with 60+ students, co-lead Google Developer Groups Debrecen, and led IBM Qiskit Fall Fest 2025 with 120+ participants. I am comfortable working across the full stack, from model training and evaluation to API design, RAG pipelines, agent systems, and deployment. Open to AI engineering roles, research collaborations, and speaking engagements.
- Represent 1000+ international students in university governance and policy discussions - Organize academic and cultural integration events with focus on professional development - Manage student union operations including budgeting, planning, and team coordination - Bridge communication between international student body and university administration Developed leadership, public speaking, and cross-cultural management skills in diverse academic environment.
Provide escalated (L2) technical support for BMW's enterprise IT infrastructure, resolving complex issues that exceed L1 capabilities - Perform root cause analysis on recurring system incidents, developing long-term solutions in collaboration with engineering teams - Create and maintain technical documentation and knowledge base articles, reducing resolution time for common issues. - Analyze incident patterns to identify systemic weaknesses and propose infrastructure improvements - Participate in on-call rotation for critical system outages, developing crisis management and communication skills in automotive industry environment. This experience with large-scale system troubleshooting informs my research into robust, fault-tolerant AI architectures.
- Build and lead technical community focused on AI/ML, quantum computing, and emerging technologies - Organize workshops and hackathons connecting 100+ local developers with industry trends and practices - Mentor developers in practical AI implementation and open-source contribution - Collaborate with Google Developers ecosystem to bring cutting-edge technical content to local community This role combines technical depth with community leadership, creating platforms for knowledge exchange in AI/ML space. Specifically focus on AI/ML and quantum computing workshops, creating bridges between academic research and practical implementation
Designing and launching an elective course titled Introduction to Quantum Computation for bachelor students • Developing weekly lab modules combining theory and hands-on coding using IBM Qiskit • Collaborating with university leadership and IBM education team • Creating visual demonstrations, assignments, and final practical exam tasks • Leading instruction and mentoring students during the semester
- Provide personalized instruction to 50+ university students and professionals in AI/ML, quantum computing, physics, and advanced mathematics - Develop customized learning approaches for complex subjects including Transformer architectures, quantum algorithms, and machine learning foundations - Maintain 4.9/5 rating through adaptive teaching methods and clear technical communication - Help students implement practical AI projects while building strong theoretical understanding - Specialize in connecting mathematical principles to modern AI/ML and quantum computing applications Teaching these subjects professionally deepens my foundational understanding, directly informing my research into AI memory systems and neural compression.