🤖 Andrzej Rębowski 🤖

Supporting a research team by developing tools and models for analyzing and recommending Computational Fluid Dynamics (CFD) simulation cases. 𝐊𝐞𝐲 𝐜𝐨𝐧𝐭𝐫𝐢𝐛𝐮𝐭𝐢𝐨𝐧𝐬: • Designed and built a recommendation system to suggest relevant CFD cases based on input parameters, model types, and performance metrics. • Developed a full web application from scratch, including both backend (data pipelines, recommendation logic, APIs) and frontend (interactive UI for visualizing and exploring recommendations). • Collaborated closely with domain experts to align the system with research needs and ensure practical usability.

Supported and optimized a large-scale legacy forecasting system and contributed to the development of a new AI commentary solution for financial reporting. Worked across the full ML lifecycle — from data preprocessing and model training to deployment, monitoring, and integration with enterprise analytics platforms. 𝐊𝐞𝐲 𝐜𝐨𝐧𝐭𝐫𝐢𝐛𝐮𝐭𝐢𝐨𝐧𝐬: • Reducced pipeline runtime from 40 minutes to 30 minutes through code refactoring and workflow optimization. • Improved data consistency and system reliability by rewriting and modularizing legacy components. • Implemented advanced prompt engineering techniques to reduce text generation time from 10 minutes to 5 minutes.

Working on a research project to develop a meta-model that can generate new neural network weights by learning and “filling in the blanks” in the noise patterns between existing models. The goal is to create diverse and performant models without retraining from scratch. 𝐌𝐚𝐢𝐧 𝐜𝐨𝐧𝐭𝐫𝐢𝐛𝐮𝐭𝐢𝐨𝐧𝐬: • Designed and implemented most of the codebase for training, inference, and evaluation. • Built a meta-model that learns the difference (noise) between trained models and generates new model weights. • Achieved accuracy within 1% of the original base model using the generated weights.

• 𝐌𝐚𝐭𝐡𝐞𝐦𝐚𝐭𝐢𝐜𝐬: Helped students learn and pass math exams at all levels, from beginner to advanced university topics, including calculus, convex programming, optimization methods, and linear algebra. • 𝐂𝐨𝐦𝐩𝐮𝐭𝐞𝐫 𝐒𝐜𝐢𝐞𝐧𝐜𝐞: Helped students with Java, C/C++, algorithms, data structures, OOP, and SQL, while also supporting them in successfully realizing their own creative projects. • 𝐌𝐚𝐜𝐡𝐢𝐧𝐞 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠: Helped students understand classical machine learning, deep learning and data science approaches. Guided them in applying these concepts to real-world problems. 𝐀𝐜𝐡𝐢𝐞𝐯𝐞𝐦𝐞𝐧𝐭𝐬: 9 students passed their exams with scores above 91%, and 2 received job offers after successful interviews.

Erasmus+ program