Mike Czech

Optimization of warehouse logistics with machine learning for time series (e.g. with XGBoost, scikit-learn, Argo, Google Cloud). Created prototypes for using computer vision in online retail (e.g. with PyTorch).

Built highly scalable machine learning pipelines and services using cloud-native architectures (e.g. with Python, Kubernetes, Terraform, Airflow, MLflow, Google Cloud, Docker, and Gitlab CI). Created and productionized regression-based models for personalized product rankings in online retail, leading to a better conversion rate in A/B tests (e.g. with XGBoost, pandas, and scikit-learn). Created and productionized NLP models for optimizing product searches (e.g. with fastText).

Developed an IoT platform that connects thousands of vehicles to services running on AWS (C++, Terraform, FreeRTOS, Lua, EC2, Docker, HiveMQ). Followed best practices like test-driven development, infrastructure as code, CI/CD, and pair programming. Integrated automatic software analyses into the CI pipeline (e.g. Clang Static Analyzer and Clang Sanitizers for C++).

Researched machine learning on graphs for program analysis and used it for automatic algorithm selection (Python, WL graph kernels, SVMs, scikit-learn, pandas) Designed and prototypically implemented novel techniques, which ease software testing of C programs by utilizing verification techniques. (Java, Concolic Testing with KLEE, CPAChecker)

Gave weekly classes on formal methods like set theory, predicate logic, petri nets, and proof strategies.

Improved the performance of 3Spot - a verification tool for concurrent systems - by implementing heuristic techniques for abstraction refinement (Scala, C++, JNI).

Gave weekly classes on advanced topics of programming in Java (e.g. concurrency, design patterns, XML parsing, and GUI development).

Simplified the presentation of data by creating a tool that generates compact PDF documents from ELO Digital Office databases (C#/.NET). Facilitated the configuration of internally used software by building a graphical XML editor (C#/.NET).

Machine Learning, Clustering, Heuristische Suchverfahren, Model-Driven Software Development. Thesis: Predicting Rankings of Software Verification Tools Using Kernels for Structured Data

Thesis: Three-Valued Model Checking of Parallel Systems with Heuristic-Guided Generation of Counterexamples