Daria Stetsenko

SOFTWARE ENGINEERING PROJECTS

Chrome Extension for Disinformation Detection.

• Developed a Chrome extension that analyzes disinformation on Twitter and BlueSky.
• Built using Python (Flask) for the backend and APIs, JavaScript for the front end, and integrated RAG, OpenAI GPT, and Tavily APIs for an agentic pipeline.
• Negotiated data access with NewsGuard to enhance detection accuracy.
• Currently used internally for surveys and experiments in the Political Department (code not publicly available).
• Public UI and release forthcoming. (link: https://bit.ly/disinformation_extension)

RESEARCH PROJECTS

Who Addresses A.I.'s Ethical and Societal Concerns?

• Investigated interdisciplinary collaboration, corporate partnerships, and academic influences on AI ethics.
• Employed Python for data processing, the Semantic Scholar API for data scraping, and fine-tuned SciBERT for embeddings extraction and classification via logistic regression.
• Utilized Ollama’s Llama 3.1 (8B model) and Llama CPP for advanced annotation and fine-tuning.
• Conducted topic modeling with BERTopic to identify thematic trends.

SOFTWARE/NLP ENGINEERING PROJECTS

• Developed English and Ukrainian versions of a stylometric tool for text mining and embedding generation.
• Evaluated and optimized machine learning algorithms (e.g., Logistic Regression, SVM, Random Forest) for diverse NLP tasks, centering on hate and offensive speech detection.
• Fine-tuned advanced transformer models (BERT, RoBERTa) for English and Ukrainian corpora, leveraging syntactic and linguistic features to enhance classification accuracy.
• Improved model performance through iterative hyperparameter tuning, loss function, in-depth error analysis, and standard deep learning best practices.
• The code repository is available via https://bit.ly/stylometrix

PROJECT MANAGEMENT EXPERIENCE:

• Negotiated with external vendors and business stakeholders to define project scope and contractual agreements.
• Utilized Confluence and Jira to document progress, maintain product backlogs, and ensure timely project delivery.
• Managed cross-functional communication channels among businesses, developers, and data analytics teams, fostering a collaborative environment.
• Analyzed web traffic and user behavior data to inform data-driven decisions and optimize website performance.
• Co-led multiple AI integration initiatives, overseeing requirements gathering, development roadmaps, and deployment of AI-powered services into the main website.