Full Stack Engineer building AI tools and web applications.

FoodChain ID

• Build AI-driven document processing tools using Python to extract chemical substances and regulatory rules from complex PDFs and Excel files. Combine rule-based logic with LLM analysis to cut manual review time and costs by ~80%.
• Lead the migration from Domino/XPages to a modern Spring Boot and Angular 18 stack. Rebuild and optimize 60+ features while bridging legacy and new systems to improve maintainability and user experience.
• Develop a multithreaded data migration pipeline using Spring, PostgreSQL, and Docker to process 2M+ entities. Expand the regulatory knowledge base by ~50% and maintain containerized deployments with multi-stage CI/CD pipelines.
• Build secure, scalable backend services for cross-company data sharing and collaboration. Implement JWT authentication, strong domain models for substances and legislation, and production monitoring to ensure reliable regulatory workflows.

AISO (ABN AMRO)

• Selected from 100+ student interviews through AISO's screening process to join a 7-person team delivering an AI adoption project for ABN AMRO.
• Served as one of the main client contacts for ABN AMRO, leading communication and delivery within the team. Diagnosed why AI tooling stays underused across a ~70-person content organisation through stakeholder interviews and an adoption survey, then designed a hands-on workshop and handover handbook to get teams actually using it.
• Contributed to AURORA, an agentic content system that turns vague requests into grounded, compliant, fully auditable drafts for a regulated banking environment. Helped build the RAG retrieval pipeline over approved articles and an evaluator that scores AI drafts against the bank's 134-KPI quality standard. This fed into a final pitch for more agentic solutions, well received by ABN AMRO's Chief Digital Officer.

Topicus

• Led a team of 5 to evaluate AI-assisted methods for extracting 1,200+ frontend strings into centralized localization files. Compared transformers, LLM-based approaches, and rule-based techniques to identify the most effective solution.
• Prototyped and benchmarked multiple language-model driven extraction pipelines, ultimately ruling them out due to accuracy and reliability constraints on real-world code. Delivered a faster, more precise regex-based solution informed by empirical evaluation.
• Built and deployed an automated translation pipeline using the DeepL API with caching and validation. Completed English and Latvian localizations, saving 100+ hours of manual engineering effort.
• Reduced localization costs by ~€7,500 and accelerated market expansion by removing extraction and translation bottlenecks. Supported future rollout of the cancer screening platform to reach more people across regions.

CODE 932

• Completed over 30 tasks for an acquisition platform using Laravel PHP, implementing screens based on design files, creating API endpoints for CRUD operations, and ensuring proper translation functionality.
• Designed and developed a booking platform prototype for a start-up client using ReactJS and Laravel, serving as a key milestone for advancing to the next TLR phase and attracting potential investors.
• Actively contributed to team meetings by reporting bugs, proposing improvements, and applying Scrum methodology to optimize task planning, time estimates, and on-time delivery.

Universiteit Twente

• Assisted over 50 students in learning programming concepts such as multithreading, OOP, and algorithmic efficiency, while supporting them in completing weekly assignments for Module 2 of the BSc Business Information Technology.
• Graded 10-15 assignments weekly, delivering detailed feedback and actionable recommendations to improve student learning outcomes.
• Evaluated and determined pass/fail outcomes for 5 student project teams' final assignments, which accounted for 50% of their module grade, based on criteria such as proper implementation of the MVC architecture.