Open-source architecture testing | Applied AI & optimization | Supply chain & operations

I work at the intersection of software architecture, applied AI and optimization, and operational systems. I enjoy turning complex technical and business problems into maintainable, testable solutions — from open-source architecture-testing frameworks to reinforcement-learning models for logistics and supply chains.

Current focus: architecture testing, software quality, applied optimization, and open-source engineering.

I contribute to several projects in the ArchUnit ecosystem, bringing architecture testing to different programming languages. The goal is simple: make architectural decisions explicit, executable, and continuously testable as software systems evolve.

Architecture-testing tooling for Java with deterministic analysis of code structure, dependencies, and architectural rules.

• Contributed across JVM and bytecode modeling, dependency extraction, selectors, architecture rules, cycle detection, layers, and JPMS-aware policies.
• Extended the project with graph analysis and visualization, reporting, CLI and build-tool integrations, security and performance validation, public documentation, and release engineering.

Architecture testing for Ruby, bringing dependency constraints and structural rules into executable test workflows.

• Recent work focuses on release readiness and project hardening, including executable quickstarts, generated API documentation, contributor and security documentation, and project metadata.
• Added extensive regression, build, installation, linting, and documentation validation to improve reliability and make the project easier to adopt and maintain.

Architecture testing for Python codebases, focused on enforceable dependency rules and maintainable application structure.

• Contribute to framework functionality, quality, and long-term project development.
• Help extend architecture testing and automated architectural constraints into the Python ecosystem.

🌐 Meet the ArchUnit team →

A reinforcement-learning approach to food-delivery logistics, using Deep Q-Networks to learn dynamic postponement decisions for order assignment and bundling.

• Evaluated on 647,395 real-world Meituan orders across 22 districts
• Achieved 5.5% lower average distance per order and 1.5 pp lower driver idle rates
• Improved on-time delivery by 4.4 pp in high-stress scenarios

Tech: Python · PyTorch · Deep Q-Networks · NumPy · Pandas

A reinforcement-learning framework for inventory and allocation decisions in omnichannel retail with resellable returns.

• Combines a Hierarchical Markov Decision Process (HMDP) with Proximal Policy Optimization (PPO)
• Achieved a 3% reduction in total costs and up to a 17% increase in service levels

Tech: Python · TensorFlow · Keras · PPO · Gym

A reinforcement-learning implementation of the classic Supply Chain Beer Game, exploring automated ordering policies under supply-chain uncertainty.

• Applied Q-Learning to optimize replenishment decisions
• Achieved a 31% reduction in total costs through an improved state-space representation

Tech: Python · NumPy · Pandas · Q-Learning

Interested in software architecture, architecture testing, applied AI, optimization, and intelligent operational systems.