References & project examples

We developed agent-based resource planning with Outlook connectivity through the Microsoft Graph API.

A LangGraph workflow drafts schedules from team capacity, working hours and project requirements. We combine structured model outputs with code-based rule checks. Outlook is connected through the Microsoft Graph API; we use LangSmith to analyse and trace agent and model workflows. Changes are confirmed before being applied.

• LLM agents
• LangGraph
• Microsoft Graph API
• Outlook
• LangSmith

Our AI moderator joins meetings, builds conversation graphs, contributes suggestions and summarises outcomes.

We combined speech-to-text (STT), text-to-speech (TTS) and LLM-based facilitation. Contributions are automatically structured as a graph. The moderator adds its own suggestions to the discussion, responds through speech and produces a final meeting summary.

• LLM
• Speech-to-Text
• Text-to-Speech
• Graph visualisation

We developed a reusable, accessible Svelte component library for the TEO v3 web interface.

UI components were developed with Svelte, TypeScript, Tailwind CSS and Web Components. The library was built and documented with Storybook; Bits UI and Melt UI extended the interface. REST APIs connect the front end to the back end. The focus was reusability, accessibility and a consistent design system.

• Svelte
• TypeScript
• Tailwind CSS
• Web Components
• Storybook

An automated test environment for iOS and Android apps was defined and connected to Jenkins.

The project analysed, defined and implemented a test environment using UIAutomation and MonkeyTalk. Jenkins handled integration; reporting and workflow organisation used Jira.

• iOS
• Android
• UIAutomation
• MonkeyTalk
• Jenkins

As part of the development team, we extended the IIoT platform with Docker microservices and TypeScript. The software collects industrial sensor data, preprocesses it at the edge and makes it available to other IT applications.

The agile development team used Azure DevOps for collaboration and development processes. We also worked on CI/CD pipelines generating SBOM files for Dependency-Track. The overall product includes C# services and an Angular frontend.

• Docker
• TypeScript
• Microservices
• Azure DevOps
• IO-Link

Screen content is captured from cameras and other image sources, then analysed with OCR and object detection.

Screen capture was implemented using USB and industrial cameras, video streaming, desktop screen capture and capture cards. OCR and YOLO object detection identify features. The UI and server communicate through TCP client/server components in C++, CLI and C#. The project also led to a contribution to the “Next Robotics” podcast.

• C++
• C#
• OCR
• YOLO
• TCP

A mobile stereo camera determines the diameter of selected tree trunks.

The prototype combines a stereo camera with two Raspberry Pis, a charging controller and a display. A 3D-printed enclosure makes the system mobile. The software calculates a disparity map and distance; a ZED stereo camera was also integrated.

• Stereo vision
• OpenCV
• Raspberry Pi
• 3D printing
• ZED camera

A CNN recognises and classifies tyres and evaluates up to 6,000 features.

A feasibility study on car-tyre detection and classification was followed by software development. Tyre images are unwrapped and corrected; a CNN recognises features. Training and augmentation of learning data were automated.

• AI
• CNN
• Computer vision
• Data augmentation

A calibrated camera and projection system maps CAD drawings to scale onto the work surface.

Calibration software was developed in C++ and OpenCV. Camera and projector are calibrated using patterns, and a camera captures the projection field. A controlled linear unit corrects height; MQTT connects the system to the digitisation process.

• C++
• OpenCV
• Camera calibration
• Augmented reality
• MQTT

AI object detection and optical measurement inspect CFRP components for CNC machines.

Development of optical defect inspection with YOLOv4 and measurement of CFRP components. The software was developed in C#, CLI and C++.

• C#
• CLI
• C++
• YOLOv4
• Computer vision