Zalán – Typescript, React, Machine learning

This cutting-edge AI platform transforms text prompts, images, and sketches into high-quality visual content, serving both creative and professional users. It features a microservice-based architecture and a self-hosted solution for image generation, leveraging fine-tuned SDXL models and FLUX-based workflows to ensure scalability and efficiency.

• Led and managed the backend development team while actively supporting the frontend team, ensuring seamless collaboration and efficient project execution to achieve milestones;
• Planned, designed, and developed the architecture, infrastructure, and key features;
• Facilitated communication and collaboration between QA, Content Management, Data Center representatives, and development teams to enhance workflow efficiency;
• Interviewed and evaluated candidates for Backend and DevOps positions to expand the technical team;
• Provided technical customer support and responded to technical inquiries on social media platforms;
• Designed and optimized a scalable microservice architecture and migrated 150,000 users to a new platform;
• Reduced infrastructure costs and enhanced image generation performance;
• Launched a public API with a credit-based system and delivered a revamped UI/UX for a better user experience.

This project is a microservice-based system designed to analyze and visualize log files generated by smart home controllers and cameras. It processes data to provide clear insights and real-time monitoring capabilities, helping users better understand the performance and status of their smart home devices. The architecture ensures scalability and efficient handling of large volumes of log data.

• Managed the end-to-end development of a comprehensive microservices solution for managing log files of smart home devices;
• Prioritized the implementation of high-security measures to safeguard sensitive data;
• Designed the system to assist and expedite the workflow of the customer support service.

A cross-platform mobile application designed for service technicians to efficiently commission, configure, and maintain heating systems. The app streamlines setup processes, provides step-by-step guidance, and enables diagnostics for a wide range of heating devices, ensuring accurate and efficient field operations.

• Designed, developed, and deployed a cross-platform mobile application for service technicians in the HVAC trade to support commissioning, repairing, and maintaining Junkers, Buderus, and Bosch heaters;
• Reconstructed the application from Apache Cordova + AngularJS to React Native using a new design system;
• Added support for new appliances and devices within the app;
• Developed in-app purchase subscription functionality;
• Built and maintained internal private React Native libraries to support app functionality and scalability.

A proctoring system designed to detect fraud during online exams, developed to support remote assessments in response to the pandemic. The platform ensures exam integrity by monitoring test-takers and identifying suspicious behaviors, enhancing the reliability of online evaluations for academic institutions.

• Led research and development efforts to analyze market trends and integrate advanced computer vision techniques for fraud detection in live video streams, audio signals, and user behavior patterns;
• Conducted thorough code reviews to enhance system performance, improve reliability, and ensure robustness in real-time monitoring environments;
• Collaborated closely with cross-functional teams to implement cutting-edge algorithms aimed at identifying cheating behaviors during online examinations;
• Focused on optimizing the detection system to operate efficiently under varying network conditions and hardware setups.

Cross-platform dog breed identifier application developed as a Bachelor’s thesis project. The app uses image recognition technology to analyze user-uploaded photos and accurately identify various dog breeds. It provides an intuitive interface for users to easily access breed information on mobile devices, combining machine learning with mobile development to deliver a smooth and educational experience.

• Designed, developed, and deployed a cross-platform mobile application working in both offline and online environments to identify dog breeds;
• Planned, designed, and developed the architecture, infrastructure, and software system;
• Researched innovative machine learning methods;
• Trained, fine-tuned, and evaluated convolutional neural networks.