𝗔𝗱𝗮𝗽𝘁𝗶𝘃𝗲 𝗕𝗶𝗱𝗱𝗶𝗻𝗴 𝗠𝗼𝗱𝗲𝗹 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁 (AWS Lambda, Sklearn, Pandas, Numpy)• Developed and deployed an adaptive bidding model for over 1 million monthly auctions, increasing revenue by 30% and profit by 15%. • Leveraged AWS Lambda and Python libraries (Sklearn, Pandas, Numpy) for scalable, real-time processing.𝗗𝗲𝗳𝗲𝗰𝘁 𝗖𝗹𝗮𝘀𝘀𝗶𝗳𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝗠𝗼𝗱𝗲𝗹𝘀 (Azure, Multi-GPU, Docker, PyTorch, ONNX)• Developed and deployed visual defect classification models with 95% accuracy in a multi-GPU, multi-node Azure environment. • Developed and integrated anomaly detection and certainty scores to ensure reliable performance.• Utilised Docker and ONNX for flexible, scalable deployment.𝗣𝗿𝗲𝗱𝗶𝗰𝘁𝗶𝘃𝗲 𝗠𝗮𝗶𝗻𝘁𝗲𝗻𝗮𝗻𝗰𝗲 𝗳𝗼𝗿 𝗛𝗲𝗮𝘃𝘆 𝗜𝗻𝗱𝘂𝘀𝘁𝗿𝘆 (Podman, PyTorch, Sklearn)• Prototyped a predictive maintenance model enabling wear prediction and fault detection, optimising operational reliability in heavy industry. • Deployed using Podman, PyTorch, and Sklearn for efficient model training and deployment.𝗔𝗻𝗼𝗺𝗮𝗹𝘆 𝗗𝗲𝘁𝗲𝗰𝘁𝗶𝗼𝗻 & 𝗧𝗿𝗲𝗻𝗱 𝗔𝗻𝗮𝗹𝘆𝘀𝗶𝘀 𝗳𝗼𝗿 𝗧𝗶𝗺𝗲 𝗦𝗲𝗿𝗶𝗲𝘀 𝗗𝗮𝘁𝗮 (Tableau, Streamlit, Sklearn, SciPy, Pandas, Numpy)• Developed an anomaly detection tool and trend analysis for the animal healthcare industry, uncovering significant historical trends linked to real-world events (COVID-19, etc.). The solution provided critical insights, improving decision-making processes.𝗖𝘂𝘀𝘁𝗼𝗺𝗲𝗿 𝗘𝗻𝗴𝗮𝗴𝗲𝗺𝗲𝗻𝘁For all projects, I served as a point of contact for customers, regularly presenting and explaining model results, ensuring they understood the impact on operational efficiency and reliability.

As an external PhD Candidate funded by the LOEWE3 program from the State of Hessen, I specialize in the development of advanced mathematical models and machine learning techniques aimed at solving complex multi-objective optimization challenges.𝗧𝗵𝗲𝗼𝗿𝗲𝘁𝗶𝗰𝗮𝗹 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁• Led the development of mathematical models parameterized over dynamic losses for manifold optimization, enhancing the performance and adaptability of multi-objective models.𝗔𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝗼𝗳 𝗠𝘂𝗹𝘁𝗶-𝗢𝗯𝗷𝗲𝗰𝘁𝗶𝘃𝗲 𝗢𝗽𝘁𝗶𝗺𝗶𝘇𝗮𝘁𝗶𝗼𝗻 (𝗠𝗢𝗢) (PyTorch, Numpy, OpenAI Gym)• Conducted extensive research and implementation of MOO techniques in diverse applications:- Image Classification: Improved cancer detection methodologies by optimizing precision and recall, utilising dynamic class weights to enhance model robustness.- Action Recommendation: Developed churn prediction models that effectively balance customer retention and profitability, driving strategic business decisions.- Reinforcement Learning: Applied MOO principles in robot control and grid world simulations, contributing to advancements in autonomous systems.𝗠𝗮𝗰𝗵𝗶𝗻𝗲 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗣𝗮𝗰𝗸𝗮𝗴𝗲 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁 (PyTorch, Numpy, Neptune)• Designed and implemented a comprehensive ML package for multi-objective models, including:- Efficient training and evaluation processes.- Data preprocessing and loading mechanisms to streamline workflows.- Tracking and logging progress, as well as publishing models in a centralized registry for enhanced accessibility.- Developed inference scripts to facilitate seamless deployment and integration of models.𝗠𝗲𝗻𝘁𝗼𝗿𝘀𝗵𝗶𝗽• Supervised multiple student theses, providing mentorship and guidance in their research endeavours.

𝗣𝗿𝗼𝗴𝗿𝗮𝗺𝗺𝗶𝗻𝗴: Developed robust solutions using Python to address complex data challenges.𝗗𝗲𝗲𝗽 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴: Trained and evaluated deep learning models for motion-tracking-based contact prediction and classification, significantly enhancing predictive accuracy.𝗗𝗮𝘁𝗮 𝗘𝘅𝗽𝗹𝗼𝗿𝗮𝘁𝗶𝗼𝗻: Conducted thorough data exploration and feature engineering for automated moment detection, optimizing model performance and reliability.𝗗𝗮𝘁𝗮 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴: Managed data preprocessing and structuring of annotation and sensor data, ensuring high-quality input for analysis.𝗖𝗼𝗻𝘁𝗶𝗻𝘂𝗼𝘂𝘀 𝗜𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗶𝗼𝗻: Implemented continuous integration practices to streamline the deployment and productisation of developed code, facilitating efficient collaboration and maintenance.

𝗣𝗿𝗼𝗴𝗿𝗮𝗺𝗺𝗶𝗻𝗴: Utilised Python to develop innovative solutions for complex data challenges.𝗔𝘂𝗱𝗶𝗼 𝗔𝗻𝗮𝗹𝘆𝘀𝗶𝘀: Implemented machine learning techniques for audio analysis and sentiment classification, enhancing the understanding of user emotions through sound.𝗔𝘂𝘁𝗼𝗺𝗮𝘁𝗲𝗱 𝗦𝘁𝗼𝗿𝘆𝘁𝗲𝗹𝗹𝗶𝗻𝗴: Developed data-driven automated storytelling systems based on annotation and sensor data, transforming raw data into engaging narratives.𝗖𝗼𝗻𝘁𝗶𝗻𝘂𝗼𝘂𝘀 𝗜𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗶𝗼𝗻: Executed continuous integration practices to streamline the deployment and productisation of developed code, ensuring efficient collaboration and system reliability.

𝗣𝗿𝗼𝗴𝗿𝗮𝗺𝗺𝗶𝗻𝗴: Leveraged Python to analyse and model correlation structures in multivariate time series data derived from biometric and empirical team sports data.𝗗𝗮𝘁𝗮 𝗔𝗻𝗮𝗹𝘆𝘀𝗶𝘀: Conducted in-depth analyses to uncover relationships and patterns within complex datasets, providing insights that inform performance optimisation and strategic decision-making.

𝗣𝗿𝗼𝗴𝗿𝗮𝗺𝗺𝗶𝗻𝗴: Utilized Java/Android and Python for diverse software development projects.𝗔𝗻𝗱𝗿𝗼𝗶𝗱 𝗔𝗽𝗽 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁: Designed and developed an Android application for collecting and processing mobile coverage data, integrating it into a central database. Responsibilities included UI design and ongoing software maintenance.𝗣𝘆𝘁𝗵𝗼𝗻 𝗣𝗿𝗼𝗴𝗿𝗮𝗺𝗺𝗶𝗻𝗴: Collected and analyzed network-related energy consumption data from System on Chips (SoCs), providing valuable insights for optimization.