Overview

Position: DAAD WISE Scholar
Institution: Berlin University of the Arts & Einstein Center for Digital Future
Supervisor: Prof. Dr. Daniel D. Hromada, Juniorprofessor for Digital Education
Duration: May 2024 - July 2024
Location: Berlin, Germany

During my three-month DAAD WISE scholarship internship, I was selected as one of 250 recipients worldwide to work on the groundbreaking Reading Primer project and fibel.digital initiative. This experience combined cutting-edge AI research with practical hardware engineering for educational impact.

Project: Personal Primer - Digital Reading Acquisition Assistant

Project Overview

The Personal Primer is a digital education artifact designed to guide students aged 5-10 in developing fundamental literacy and numeracy skills. It functions as a user-friendly learning device—specifically, a Digital Reading Acquisition Assistant (DRAA)—that embraces the innovative concept of ‘human-machine peer learning.’

My Role and Contributions

Speech Recognition Development

  • Developed and programmed sophisticated speech recognition models, particularly DeepSpeech and Whisper, for Human-Machine Peer Learning (HMPL) modules
  • Deployed audio-textual contents from Book 1 of the Indian Fable collection, Panchatantra, on Raspberry Pi-driven Personal Primer Prototypes
  • Created a “personalized narration module”—a complex system integrating language models, speech recognition, and adaptive learning systems

Technical Challenges Solved

  • Constructed Automatic Speech Recognition (ASR) system capable of handling imperfect speech input—critical for young learners
  • Personalized the Primer system to adapt to individual pupils’ learning patterns and reading progress
  • Optimized system performance for Raspberry Pi constraints while maintaining educational effectiveness

Hardware and Software Engineering

Prototype Development

  • Engineered the prototype ‘Fibel 4’, significantly enhancing both hardware and software capabilities
  • Improved soldering techniques and acquired proficiency in laser-cutting and 3D printing (both filament and resin methods)
  • Mastered Raspberry Pi and Raspbian Linux development environment

Technical Skills Acquired

  • I2C sensors integration for interactive hardware components
  • Bash scripting and vim proficiency for system administration
  • Vector databases (Chroma) for efficient content retrieval
  • Exercise-specific language models with audio-context support for speech recognition

Sustainability and Impact

Sustainable Design

  • Solar energy integration—the device utilizes renewable energy, making it cost-effective and suitable for widespread adoption
  • Aligned with Sustainable Development Goals (SDGs) for accessible education technology
  • Optimized for resource-constrained environments while discouraging prolonged usage

Research Integration

  • Gained proficiency in udk.ai’s personalized NLP models integral to the project’s success
  • Demonstrated practical application of advanced AI in educational technology
  • Built complete functional prototype in just three months, showcasing rapid learning and adaptation

Skills Developed

Technical Competencies

  • Speech Recognition Systems: DeepSpeech, Whisper, ASR optimization
  • Hardware Engineering: Raspberry Pi, I2C sensors, 3D printing, laser cutting
  • Software Development: Python applications, bash scripting, Linux system administration
  • AI/ML Integration: Language models, vector databases, adaptive learning systems

Research and Development

  • Rapid Prototyping: From concept to functional device in three months
  • Cross-disciplinary Integration: Combining AI, hardware, and educational theory
  • Sustainability Focus: Designing for global accessibility and environmental responsibility

Impact and Recognition

This internship solidified my understanding of my potential for a brilliant computer science research academic career. The experience demonstrated my technical abilities, proactive attitude, and collaborative spirit in an international research environment.

The project showcased how advanced AI technologies can be made accessible and practical for educational applications, particularly in resource-constrained environments where traditional educational technology might not be feasible.

Future Applications

The skills and experience gained during this internship directly contributed to my subsequent success in production NLP roles, demonstrating the value of combining theoretical research with practical, hardware-constrained implementation challenges.