Abdelrahman Abozied

• Contributed to the Syntea Department, enhancing Syntea, an AI-powered learning assistant serving 130,000+ students with personalized, interactive learning support. • Worked on the SynteaOS team, developing the chatbot designer tool, a FastAPI-based platform enabling developers and professors to create custom AI chatbots using various LLM providers for tailored educational experiences.

• Developed AI services to enhance chatbot interactions, integrating with LLMs via function calling for smarter responses and expanded capabilities. • Owned the design & implementation of a FastAPI service using the OpenAI Assistant API, enabling seamless multi-assistant integration. Built a scalable, dynamic system that manages conversation history, creates assistants on the fly, and empowers non-technical users to configure AI workflows while enabling intelligent specialist escalation when needed.

Working Student – TUM.ai Industry Program Phase 4 Participated in a 12-week AI consulting program, collaborating with industry partners to develop a fraud detection proof-of-concept. Worked with domain experts, refined feature engineering strategies, and integrated diverse financial data sources. Built a PySpark-based pipeline, conducted EDA, and developed ML models (SVM, Random Forest, etc.). Delivered a working fraud detection model with clear documentation and stakeholder presentations.

Translated business requirements into AI and machine learning solutions. Utilized NLP techniques to collect and filter data, extracting valuable insights. Worked on large-scale, real-time data and analytics projects, employing advanced machine learning models. Designed and implemented a new database by scraping data from various sources, conducting analysis, and preparing it for use. Engineered cost-efficient alternatives to a paid API service, resulting in significant cost savings for the company.

Contributed to the Editing Large Language Model Predictions with Natural Language Feedback project, working under the supervision of Prof. Derry Wijaya and mentorship of Afra Feyza Akyürek, a Ph.D. student.Conducted research on improving the accuracy of OpenAI GPT-3 code-davinci-002 predictions on GSM8K and MBPP datasets, using natural language feedback. Created and implemented new ideas to enhance the accuracy of the project's predictions and generate high-quality prompts.

Acquired a comprehensive understanding of AI and Machine Learning fundamentals through extensive training and hands-on projects. Built machine learning models using the scikit-learn library and gained hands-on experience. Designed and implemented deep learning models using TensorFlow and Keras, solidifying my knowledge of deep learning algorithms and techniques.

Acquired knowledge on generating innovative ideas through design thinking and mind-mapping techniques. Improved entrepreneurial skills by gaining exposure to the fundamental principles of entrepreneurship and interactive learning opportunities to apply essential tools to turn ideas into a business. Demonstrated leadership abilities by managing team tasks and overseeing performance with mentor guidance.

GPA: 3.64 / 4 (1.3 German Scale) Grade: Excellent Thesis: Emotion Recognition using Artificial Intelligence Solutions. (A* Excellent)