🎓 Computational Social Scientist & ML Engineer | Human-Centered AI

I build and study AI systems as tools for decision-making, education, and social inquiry. My background spans philosophy, economics, and quantitative social science, and my work sits at the intersection of machine learning, human judgment, and ethical evaluation.

With 8+ years of experience across research, consulting, and product development, I’ve led large-scale AI deployments, published peer-reviewed research, and built generative AI systems that support evidence-based feedback and learning. I’m especially interested in how uncertainty, bias, and evaluation shape the real-world use of AI.

📰 Who Counts as Working Class? — Computational Social Science (2025–Present)
Large-scale media analysis of how “working class” is represented across 40,000+ articles (1980–2024) using NLP and generative AI. Conducted in collaboration with a professor at Columbia University.

✏️ WriteWise — Human–AI Evaluation in Education (2025–Present)
An AI-powered formative feedback system for writing assessment. Uses agentic LLM workflows grounded in learning science, with multiple feedback strategies (Glow & Grow, Rubric-Based, Error Spotting).
→ Actively developing evaluation pipelines with Columbia University practicum students to measure pedagogical effectiveness and user trust.

🤖 Writing Feedback with Llama — Open-Source LLM Fine-Tuning (2024–Present)
Research on fine-tuning open-source LLMs for grading and justification, focusing on improving explainability and alignment with human judgment.

What LLM Storytelling Tells Us About LLMs — Experimental NLP, 2025
Experimental study of how narrative generation reveals latent values and biases in GPT-4.1.

Bayesian Time Series with Stan — Bayesian Forecasting, 2024
Bayesian ARIMA with full posterior uncertainty propagation and comparison to frequentist forecasting.

Cross-Voting in Mexico’s 2024 Election — Political Data Science, 2024
Interactive Python + Leaflet analysis of split-ticket voting in a historic national election.

Twitter Sentiment & Urban Well-Being — Landscape and Urban Planning (Journal), 2019
Peer-reviewed study of 3.3M geolocated tweets showing how social media sentiment can serve as a proxy for urban well-being.
→ 📈 228 citations on Google Scholar.

• Built an NLP system to process years of unrecorded inmate-submitted data, uncovering 1M+ days of missed sentence reductions and saving $100M → TEDx Talk (2022)

• United Nations collaborations with UNODC, UNDP, and ILO on trafficking, gender equity, and financial inclusion in India.

Using a Glacier Website to Promote Action and Build Community — Climate, Capitalism, and Communities

Indebted to Work: Bondage in Brick Kilns — Palgrave Handbook of Bondage and Human Rights

Research on Food Deserts — Statistical Analysis

• Programming: Python, R, Stan
• Machine Learning: Scikit-learn, PyTorch (model training & fine-tuning)
• NLP & GenAI: Hugging Face Transformers, LangChain, Agentic Workflows, RAG, LLM Evaluation Pipelines
• Statistics: Bayesian Hierarchical Models, GLMs, Experiment Design, Time-Series Forecasting

• M.A. Quantitative Methods in the Social Sciences, Columbia University
• M.A. Food & Resource Economics, University of British Columbia
• B.A. Economics & Philosophy (Honors), UC Santa Barbara

• LinkedIn: https://www.linkedin.com/in/laurauguc/
• GitHub: You’re here 🙂

I’m always interested in collaborations on AI as a tool for scientific reasoning, evaluation, and human judgment—especially in education, social systems, and public decision-making.