Berlin, Berlin, Germany
I'm a Solutions Engineer III at GitHub, supporting corporate customers across EMEA. I help prospects and existing customers understand, evaluate, and get the most out of the GitHub platform by running live demonstrations of GitHub Enterprise, answering technical and product questions, clarifying capabilities and integration paths, and partnering with sales, support, services, and product teams to build strategies that expand how organizations adopt GitHub at scale. What I enjoy most is the translation work: turning a customer's real-world challenges into a clear technical story, surfacing their feedback inside GitHub, and advocating for their interests so the platform genuinely fits how they build software. My foundation is in machine learning and applied AI. I have hands-on experience with natural language processing and large language models across domains like sentiment analysis, text generation, and recommendation, which gives me a strong grounding for the increasingly AI-driven conversations customers want to have today. I hold a master's degree in Cognitive Systems from the University of Potsdam, where I studied the intersection of language, learning, and reasoning. Alongside my work at GitHub, I freelance for O'Reilly, developing on-demand courses and facilitating live webinars on NLP and LLMs. I'm passionate about sharing knowledge with the wider AI community as a public speaker, a blogger, and a diversity, equity, and inclusion advocate.
The Solutions Engineering team helps prospects and existing customers successfully understand, evaluate, pilot, and leverage the GitHub platform in coordination with a sales process. We achieve this by knowledge-sharing, collaborating on Q&A, writing documentation, and providing the sales team with information that will answer frequently asked questions, for example: - Provide live demonstrations of GitHub Enterprise to prospective customers - Build joint customer strategies with enterprise sales, support, services and product management to expand current usage of GitHub Enterprise across an enterprise at an executive level - Answer GitHub product questions and help customers and prospects to take full advantage of the GitHub platform - Answer GitHub technical questions and help customers to integrate GitHub with their existing software and processes - Clarify technical capabilities, solutions, and limitations of the GitHub platform - Learn about customers' successes and challenges, share customer feedback within GitHub, and advocate for customer interests - Promote the adoption of GitHub within customer organizations - Answer pre-purchase security questions about GitHub.com and GitHub Enterprise
I created and teach the online courses: - On Demand: Natural Language Processing Fundamentals: Techniques, Tools, and Applications: https://learning.oreilly.com/course/natural-language-processing/0790145481825/ - Webinar: Running Open Source LLMs Locally (Ollama) - Webinar: Local LLMs Made Easy - Using UI-Based Frameworks Jan AI and GPT4ALL
API development/containerization/deployment/monitoring of the following projects: - Decrease of support tickets created by users by 9% by implementing a Q&A chatbot with LLMs and RAG end-to-end. - Increase of number of in-app payments by 8% and in-app net revenue by 14% for users in Great Britain by improvements of the GNN recommendation system. - Automated detection through image classification of hundreds of male users daily who are avoiding the paygate by creating accounts as female users. - Automated detection and flagging of suspected minors on the app by image classification. - Built multiple demos and proofs of concept of innovative features for stakeholders using LLMs and other machine learning tools, e.g. for customized content generation and user search experiences.
Reimplementation of Google Cloud Natural Language API for a social media analyser - Decrease monthly usage cost from €450.000 (Google’s API) to €30.000 (our API) by combining multiple open source models on GCP, tailoring the extraction of focus words for the client’s needs. - Ensured dynamic scaling from 0 to 1.5 million processed requests per day with a cache mechanism to optimize performance and cost. Generating business insights by analysing recorded customer calls of a leading media distributor - Set up a labelling environment on Label Studio and wrote a labelling manual. - Conducted customer training sessions and workshops on labelling speech data, guiding clients through the process through iterative feedback. - Decreased the transcription error rate from 53% to 33% by fine tuning a transcription model (ASR) on the labelled data. Digitalising scans of newspapers from a newspaper distributor - Established a labelling environment, conducted client workshops on labelling newspaper scans and guided the labellers through the process through iterative feedback. - Conducted a workshop for clients on labelling newspaper scans and guided the labellers through the process through iterative feedback. - Improved newspaper accessibility by fine tuning a model on the labelled data that digitalises them. (Selected projects)
- Increase visibility for ML6 and regularly share insights at various events and conferences. Known for communicating complex technical concepts in an engaging and accessible way. - Advocate for empowerment and DEI by giving workshops such as the one from Google’s #IAmRemarkable initiative inside and outside ML6.
1-month project of the Flanders Research Information Space (FRIS) for the Belgian government. My team and I developed a service that enriches the metadata of the search engine with new keywords.