PDF Expert in Munich

• Governed multi-agent orchestration for regulated, EU-based companies – self-hostable, compliant with the EU AI Act and GDPR („by design“), BYOM (own models/GPU).
• Two-gate governance: agent deliberation + mandatory human approval, full signed audit trail; graduated autonomy model („internal → autonomous per skill“).
• Verified knowledge graph („Company Brain“) with source evidence for every answer; own orchestration framework (Virtual Team Framework).
• Industry solutions for financial services: compliance monitoring, invoice and contract review; hands-on development with LLMs (including Anthropic/Claude), agentic workflows, RAG.
• Building the governance-focused multi-agent platform TeamIntel (see AI reference projects).

• Set up GUI test framework for testers and developers to easily create GUI tests
• Implemented automated GUI tests with QF-Test for stock trading software and PDF file comparisons
• Created comparison runs between different versions

Used technologies: QF-Test, Java, Jython

• Responsible project manager for complex international infrastructure projects
• Coordinator and decision-maker for IT architects in network infrastructure and software with over 100,000 users
• End-to-end project management
• Launch and implementation of initiatives
• Project management of international infrastructure projects with C-level reporting
• Leading and coordinating IT architects
• Technical project management for merger and cloud transformation
• Delivery of international software and hardware rollouts
• Implementation of IT security projects and compliance
• Work with key technologies (e.g. 5G/6G standard)
• Support of complex defense and aerospace projects (e.g. Ariane 6, Eurodrone)
• Building the healthcare and life sciences business unit
• Creating and leading internal employee trainings

• Led the project to implement the Customer Information Tool for the Microsoft AI Tour EMEA
• Directed the development of an event platform with sponsors, speakers, and agenda
• Conducted a compliance audit to ensure adherence to Microsoft corporate policies
• Coordinated vendors, dependencies, and deliverables across multiple European tour stops and trade show appearances

• Development of an accessible, responsive website based on Figma designs for a major sports event in compliance with WCAG, BITV, and EN 301 549
• Technical planning and setup of the tool stack with Vite, Playwright, Docker, and GitLab CI
• Worked on two existing projects (Vue.js & Nuxt), with a focus on bug fixing, accessibility audits, and new feature development
• Collaboration with design and QA to consistently implement accessible UI patterns
• Consulting on improving existing frontend architectures with regard to accessibility, performance, and maintainability

• Participated in the Riyadh Public Transport (KAPT) project introducing public transportation in Riyadh, supporting multimodal journeys combining bus, metro, car on demand.
• Worked via Thiqah and partnered with Andersen to refactor the mobile MaaS application DARB into a new microservices architecture under SCRUM framework, successfully completing Phase 1.
• Facilitated a team of 40 developers across Android, iOS, Web, Java, DevOps, architecture, analysis, and QA.
• Moderated Scrum events and communicated with the client, proactively reporting on project deadlines, scope, and challenges.
• Coordinated cross-team efforts between Thiqah specialists and GPS development teams, managing Jira and Azure DevOps trackers.
• Supported and improved Scrum processes throughout the project.
• Due to Thiqah’s takeover by Elm, subsequent phases were handled entirely in Saudi Arabia with GPS providing IT consulting.
• Tools: Azure DevOps, Jira, Confluence, Draw.IO, ChatGPT

• Set up the new digitization strategy.
• Using Camunda as a process engine, a new way of handling batch and dialog tasks is created.
• In close cooperation with the business department, the individual processes are created and put into production.
• Forms for controlling manual interventions, e.g. when adjusting applications, are implemented.
• Gradual migration of old code from IBM WebSphere to JBoss.
• Skills: Arquillian, Camunda, Initiative, Empathy, Hibernate, Jackson, Java/JEE, JBoss, Jersey, JSON, JUnit, Openness to criticism, Willingness to learn, REST, Teamwork, WebServices.

• Owning the entire hiring funnel: defining the requirements, conducting the interview process and aligning with the talent acquisition team for hiring capacity
• Planning and monitoring resources: building an initial business plan in close coordination with the finance team and revisiting this on a monthly basis to control the achievement
• Alignment with the core team: setting up rules and processes together with the team leads of the core team in order to ensure a value driven collaboration
• Reporting to C-Level: weekly updates with CEO & COO with presentations and status reports in order to show advancements and challenges
• Creating incentive schemes: creating new incentive schemes for the different team members in order to align them with the common goal
• Coordinating ad campaign with the paid marketing team
• Coordinating webinars which highlighted new aspects of the products with the marketing team
• Coordinating specific dashboards to monitor performance with the data team

• Architected and implemented an enterprise-grade AI Agent Platform leveraging Retrieval Augmented Generation (RAG) architecture to enhance clinical data insights.
• Established robust CI/CD pipelines for LLM applications using CDK and Jenkins, significantly reducing deployment times.
• Implemented comprehensive observability solutions that increased agent reliability across pharmaceutical environments.
• Designed scalable AI workflows with advanced orchestration that optimized context handling for enterprise data sources.
• Technologies: AI Agents (LangChain, LangGraph, Bedrock, Smolagents, Streamlit); LLM Operations (Tracing, Testing, Evaluation, LangSmith, LangFuse); Infrastructure-As-Code (AWS CDK, Terraform, Typescript, Jenkins); Vectors, Embeddings, RAG (OpenSearch, pgvector, PDF Extraction)

• Designed and implemented an advanced, agentic RAG pipeline using LangChain and LangGraph for structured data extraction from PDFs, utilizing tools, state management, and OpenAI LLMs (GPT-4) to improve accuracy and handle complex document structures.
• Developed a Google AI agent for extraction of structured information from PDF documents and deployed the agent on Vertex AI.
• Architected data pipelines using Azure Data Factory and Databricks to ingest data from Azure Blob Storage, process it with PySpark, and load it into Azure SQL Database via Linked Services.
• Containerized AI agents and services using Docker for consistent local development and deployment.
• Utilized PySpark and Dask for database querying in coordination with Azure Blob Storage and Document Storage.
• Created a ReAct agent that extracts structured data from PDF documents using tools and integrating Azure Document Intelligence.
• Contributed to the CPO invoices validation check project using Databricks to find existing CDRs and calculate total valid costs.
• Developed a conversational AI agent (chatbot) with a FastAPI backend, integrating RAG for precise tariff extraction and deployed the service using Azure Container Apps.
• Tools used: Azure, Azure OpenAI, Azure Document Intelligence, Azure Blob Storage, Google ADK, Google Cloud, Vertex AI, Gemini, Databricks, LangChain, LlamaIndex Ollama, Docker, PySpark, Azure SQL, Azure Data Factory, Azure AI Agent, Microsoft SQL Server

• Support and consulting for implementing ASF print components used to create PDF forms
• Technologies: DCF, ASF, DMS, SQL
• Development
• Documentation
• Program adjustments for ASF->FDVG migration

As part of an in-house development project, a pipeline for automated lead generation was created as the second stage of a preceding system for project monitoring. Based on a list of freelance project URLs (freelancermap), qualified lead records are generated, including company, contact person, official and personal email address, and the appropriate form of address (informal/formal). The technologies used were Python (openpyxl, requests) as well as an LLM agent workflow in VS Code (GitHub Copilot Chat) with a custom slash command and extensive rule set; search APIs (Serper.dev) are connected for research. My tasks included the full concept and development. The core is a rule set of around 480 lines that guides the LLM agent deterministically through extraction, website and email lookup, duplicate matching (against existing provider lists), and the creation of a structured JSON output. Deterministic steps (Excel matching, web/email search, pattern derivation) were moved into Python helper scripts. Other requirements included validating the results through versioned blind runs against a reference, iterative refinement of the rule set, and a swappable search provider layer for cost and stability reasons.