My work history

• Conducting an in-depth review of state-of-the-art multimodal methods.
• Collecting large-scale dataset for multimodal recommendation.
• Developing a multimodal recommender system.

• Investigated how our previously developed single-branch recommenders perform in comparison to their multi-branch counterparts.

• Developed an efficient multimodal (text, image, audio, ...) recommender systems for missing modality and cold-start scenarios,
• and investigated adversarial training to unlearn multiple attributes simultaneously from VAEs.

• Developed a monitoring dashboard to optimize container deployments on GCE, Azure, and AWS.
• Leveraged Python (back end) and JavaScript (front end) to build a user-friendly interface for production.

• Utilized adversarial training to improve user-side fairness in recommender systems,
• researched on how to describe users and items in terms of prototypes,
• and built EmoMTB, a music exploration interface.

• Academic Research:
  • Explored and developed methods for debiasing recommender systems
• Music Tower Blocks (EmoMTB) | JKU LIT @ Ars Electronica:
  • Designed, implemented and improved EmoMTB, our large-scale music exploration interface
  • Implemented controller interface for mobile phones
  • Handled communication between controller and exploration interface
  • Presented our interface during the festival

K – JKU’s Interactive Robocar | JKU LIT @ Ars Electronica Festival

• Utilized Python, Robot Operating System (ROS) framework, and SLAM to provide localization of our autonomous car.
• Used OpenPose to detect hand gesture commands to control the car.
• Presented project to visitors during the festival

Teaching Assistant for the courses Programming in Python 1 & 2 of the Institute of Machine Learning.

• Developed software solutions for automation systems
• In-house and at customer installation of software solutions
• Developed various tools to ease order processing