Data Scientist

Develop unit tests in Kotlin to validate the functionality of an AI-driven story generation system.
Test and refine prompts to ensure creative consistency and optimal performance of AI models in producing engaging narratives.
Operate within a live deployment environment, simulating production conditions to troubleshoot and enhance system performance.

Developed over 4 clean and reusable packages in Pytorch for academic purposes by creating templates and streamlining submission processes.
Managed virtual machines and created server and Docker environments to support deep learning projects, ensuring a functional and expandable infrastructure.
Operated within the 'Speech, Language, and Deep Learning' Lab, researched automatic speech recognition using Whisper, HuBERT, and Wav2vec models, performed speech and text alignment and wave restoration with the U-Net model.

Delivered actionable insights from large datasets of unstructured reviews to management, facilitating data-driven decision-making and enhancing strategic product development based on customer feedback.
Conducted the research, development, and implementation of advanced NLP systems, including the fine-tuning and error analysis of BERT models for text classification and T5 for question-answering tasks.
Leveraged cloud environments with AWS S3 and EC2 to efficiently store and process large datasets, enabling seamless fine-tuning of NLP models.
Performed within a multidisciplinary team of Python and R developers, data analysts, and statisticians under a product manager’s guidance to achieve project deliverables and ensure compliance with technical standards.

Fine-tuned the DistilBERT model from HuggingFace, increasing sentiment classification accuracy from 47% to 86.20% across three categories (positive, negative, and neutral).
Implemented text classification models, including Random Forest, CatBoost, XGBoost, RoBERTa, and DistilBERT, utilizing Scikit-learn and Transformers packages to enhance predictive accuracy and performance.
Delivered a comprehensive presentation of findings to students and lecturers, communicating technical details and insights.
Collaborated closely with team members and operated under the guidance of a senior data scientist, efficiently sharing responsibilities and contributing to team success.