Zilong Zhao

• Evaluation and use of cloud-based data platforms (Databricks, Snowflake, Azure Synapse) for the development of scalable ML solutions for industrial IoT applications, including cost estimation as a basis for decision-making. • Establishment and leadership of an international, cross-organizational team for the development of an Industrial Machine Learning-as-a-Service solution using Snowflake and TensorFlow, including the implementation of a data mesh for industrial IoT use cases.

Digital Automation & Industrial IoT Solutions • Development and maintenance of an on-premise data platform for data processing using Prefect and MLOps with MLflow on Kubernetes as an initiative for industrial IoT applications. • Cross-functional Collaboration: Consulted external clients to assess data acquisition, processing, and modeling needs, designing tailored data solutions aligned with business goals

Digital Automation & Industrial IoT Solutions • Built & deployed WebApps using React.js, Material-UI, Chart.js, and Kubernetes on Azure. • Engineered a real-time industrial PC monitoring tool with Python, Grafana, and OPC-UA. • Collaboration with Microsoft and IoT stakeholders, led the development of a data pipeline for processing high-frequency sensor data from industrial transport systems via OPC-UA server on automation PCs, enabling real-time anomaly detection with Azure

Industry 4.0 / Smart factory • Recommended data science tools, storage solutions, and scalable pipelines for IIoT and analytics applications. • Developed interactive visualizations with Bokeh, HoloViews, Google Charts, and D3.js; built dynamic dashboards using Dash, Flask, PySpark, and Spark-submit. • Implemented advanced analytics, including time series forecasting (ARIMA, R forecast), anomaly detection (TensorFlow Autoencoders), and NLP (NLTK).

- Image data manipulation with PySpark and HDFS - Deep learning modelling - Compiled TensorFlow r1.3 with GPU and Hadoop support on a Cent OS server - Deployed TensorFlowOnSpark - Deployed distributed deep neural network training, using Spark Streaming - Hacked TensorFlowOnSpark for asynchronous training deep neural network with distributed TensorFlow and one GPU in a YARN cluster - Developed Mr. Character, an expert in hand-written letters and digits

statistische Analysis, Datenaufbereitung, Datenvisualisierung, Datenmodellierung, maschinelles Lernen

- Numerical analysis and mathematical modelling of liver inflammation data - Leading a project for teach4TU - applying concepts of mathematics to problems in engineering - Attending a workshop on multiscale modeling of cellular and molecular biotechnology at Max-Planck-Institute for Dynamics of Complex Technical Systems, Magdeburg

- Worked on long-term measurements and mathematical analysis of mechanical vibrations of the new X-ray Free-Electron Laser (SwissFEL), and on short-term vibration measurements of the Swiss Light Source (SLS) synchrotron, processing a large volume of ASCII data (about 1 Gbyte/day) using IDL programming language - Performing mathematical research on vibration data, summarizing analytical results, writing reports

mit Schwerpunkt Technik / Naturwissenschaften