Lukas Famula

The project aims to develop a specialised AI-powered system that enables automatic transcriptions of audio files for podcasts. By implementing advanced deep learning algorithms tailored to the challenges of podcast transcription, high accuracy and reliability in transcription is ensured. My role is to implement this AI system to enable podcast creators to effectively manage and market their content.

On this project for the heavy metal band Rigorious, I was responsible for connecting to the headless CMS Sanity.io, creating the data models for the content structure, and connecting to external data sources such as Bandsintown and Spotify.
I also acted as an interface between the band and the design team to ensure that the band's wishes and requirements were met.

As part of the further development of the cash flow and liquidity planning system, I focussed on the development of additional modules in order to integrate further data groups into the system. I also worked on the quality assurance of the existing code in order to improve the robustness and reliability of the system by eliminating potential weaknesses.

The project to redesign Geschenkzeitung.de involved the development of a modern website with a fresh logo and an improved user experience. As part of the development team, my role was to connect external interfaces to ensure the integration of services and features. By implementing these interfaces, a seamless connection between the website and external services was created, which helped Geschenkzeitung.de to provide an improved service to its customers.

As part of the project to develop an advanced AI system for the automatic detection and classification of hazards in fluoroscopy facilities, I was responsible at various levels. I led the project planning and control and monitored the progress of the implementation of image data pre-processing techniques to create an optimal basis for AI analysis. In addition, I planned and realised the data acquisition and designed customised AI models tailored to the specific requirements of the plants. Furthermore, I conducted the training and validation of the AI models and designed and built embedded hardware architectures to efficiently integrate the AI algorithms. Finally, I implemented the user interface to enable user-friendly interaction with the system.

Working closely with customers in the automotive industry, we developed custom AI systems for predictive maintenance of casting processes. My role involved pre-processing and analysing image data to enable accurate predictions and to ensure that the AI models were trained with high quality data. I was also responsible for the implementation and run-time optimisation of the AI systems, with a particular focus on efficiency to ensure seamless integration into our customers' production processes.

As part of a project to develop state-of-the-art microarchitecture applications in the context of a merchandise management system, my focus was on implementing an anti-corruption layer as an interface between the individual systems. This layer ensured secure and reliable data transfer, regardless of the different data formats and protocols of the various systems. I was also responsible for the connection to an SAP system using a REST API to enable smooth communication and data transfer. These measures maximised the flexibility, scalability and efficiency of the overall system and ensured seamless integration into existing SAP systems.