Working under the supervision of Prof. Stephan Mandt on Generative models.

Worked on improving the AI assistant in Adobe acrobat

- Conducted cutting-edge research in sound-guided image manipulation and generation projects as well as audio identification and re-synthesis tasks.- Leveraged Diffusion-based models, designed and implemented various adapter architectures for efficient audio conditioning.- Explored and tested various image inversion techniques, pinpointing the best method for image identity preservation and editing ease.Fine-tuned various audio encoder models such as AudioCLIP, CLAP and fine-tuned baseline models on custom dataset.

Conducted research under supervision of Prof. Aykut Erdem and Prof. Erkut Erdem. Worked on Denoising Diffusion models in multi-modal domain.

- Conducted extensive literature review on grammatical error correction for multiple languages, to see the best options for Turkish- Training and evaluating multiple baselines and SOTA models for grammatical error connection.

- Assisted in organizing and coordinating Hackathons in the Team. - Helped in arranging and directing workshops and mock interviews with various companies.

- Worked on deep learning-based solutions for camera calibration, tailored to thecompany-specific scene geometry.- Developed fully-automatic camera calibration pipeline.

Created dynamic pdf reports using Node.js, Puppeteer and Handlebars for customer analytics report.