Peter Zaki

- Led product direction and execution, delivering the initial PoC and MVP to the first pilot. - Collaborated with stakeholders to align business vision and product requirements. - Architected and deployed the cloud infrastructure, including backend services and a secure real-time chat server. - Implemented secure authentication, user provisioning, and real-time communication with end-to-end encryption. - Designed and implemented the frontend and UX. - Designed a Retrieval-Augmented Generation (RAG) architecture to enable future ML-powered features.

- Thesis title: Hybrid Deep Reinforcement Learning Algorithms for Vendor-Managed Inventory Optimization in ATM Cash Management. - Developed a novel deep reinforcement learning framework by integrating the Vendor-Managed Inventory (VMI) model and hybrid RL algorithms. - Preprocessed and prepared a time-series dataset of daily ATM withdrawal records from over 2,000 ATMs across two years. - Designed and implemented feature engineering workflows including statistical, and temporal features to improve the state representation for the RL agents. - Utilized forecasting model with Bayesian optimization to predict future demand and fine-tune model performance.

- Development of a reinforcement learning model to enhance the ATM restocking process. - Testing the models' performance compared to the existing system. - Documentation of the entire development process and findings.

- Design and automate diverse test scenarios to evaluate system functionality, identifying and resolving issues. - Continuously improve testing methodologies, document findings, and deliver comprehensive reports with insights and recommendations.

- Adapted Simulink models to generate C code using B&R automation studio target for Simulink. - Configured B&R real time automation PC to execute software and adapt its interfaces. - Configured and connected cables and plugs ensuring proper setup and functionality of hardware systems. - System testing.

The Manage and More program of the UnternehmerTUM aims to qualify entrepreneurial talents for their ongoing professional career. Each semester 20 particularly motivated students are selected for this 18-month program. The scholarship holders develop themselves on a personal and professional level throughout innovative project work, workshops, coachings and mentoring, in order to take future responsibilities as founders or executives.

Focus on Automation and Robotics / Machine Learning