Cologne, North Rhine-Westphalia, Germany
Prototyping Environment Development of a platform for rapid experimentation and demonstration of new AI capabilities. Selected highlights: • Full end-to-end development of the frontend (Flask) and backend (FastAPI), including container deployment and monitoring. • Transformation from an internal testing site into a productive platform with a backend now actively used in production. • Implementation of multiple prototypes such as Translate Form, Form Creation from Text, Form Creation from PDF, Nova, and the Form Editing Bot. • Direct applicability for customers to test and experience new AI features at an early stage. Agentic ChatBot Development and go-live of Nova, a multimodal, multi-tool agentic chatbot. Selected highlights: • Answers general questions about smapOne, smaps, and the API; creates and analyzes smaps; generates and improves formulas. • Built on a robust architecture using FastAPI and LangGraph. • Multi-layered security system for anonymization, validation, and protection of sensitive data. • RAG architecture leveraging FAISS and embeddings for performant and reliable knowledge retrieval. • Deployed in production with containerized operation, CI/CD pipelines, and comprehensive monitoring. Training and Deployment Infrastructure for AI Models Design and implementation of a flexible infrastructure for training, deploying, and operating language models in production. Selected highlights: • Training and evaluation of open-source / open-weights models for smapOne-specific use cases. • Setup of a scalable environment enabling rapid model deployment. • Seamless integration into the smapOne backend, allowing direct access to in-house models. • Strengthening independence from external providers and establishing a future-ready, self-sufficient AI infrastructure.
• Innovated in AI Application: Spearheaded the integration of Large Language Models (LLMs) for automating data-gathering form creation on our no-code platform, a pioneering move in our industry. • Prototype Development & Client Engagement: Developed an initial prototype and led its evolution through internal and client-involved testing, culminating in the creation of over 1,700 forms, with about 100 active testing clients. • System Integration & API Development: Confronted code replication challenges during our technology's platform integration. Successfully overcame these by developing a production-ready API, greatly enhancing form creation and editing processes and streamlining frontend integration. • Continuous Improvement & Recognition: Actively fine-tuning our LLMs to better meet specific client needs, reinforcing our AI innovation leadership. Awarded the 'Outstanding Impact Award' for these significant advancements and contributions to the company.
• Drove the training and implementation of ML-driven models using PyTorch, fortifying our product segmentation approach. • Conducted a comprehensive exploratory analysis on competitor products, examining ingredients, classes and functions. This in-depth study furnished the team with valuable insights, providing a baseline for subsequent detailed analyses. • Independently identified and addressed a bottleneck involving manual tasks, creating an automation tool that substantially reduced costs and boosted operational efficiency. • Collaborated intensively with data engineers, further refining dataset mapping algorithms to enhance operational outcomes, which resulted in a 60% reduction in processing time.
• Data Science: Building data pipelines and their respective dashboards to visualise and analyse data • Digitalisation / Automatisation with Google Apps Script • Web development • Support the planning as well as the customer service team
• Coating powder experiments on fluidized bed • Maintenance work on fluidized bed system • Shear stress measurements • Lignin Milling