Decision-safe evaluation + Streamlit dashboard for AI vs Human vs Post-Edited AI text detection. Generates a reliability report card (Accuracy, Macro F1, ECE, Brier), calibration plots, confidence histograms, and a coverage-vs-performance abstention curve. Recommends an operating threshold for human-review routing.

How Chain-of-Thought Budgets Induce Overconfidence in LLMs. Investigating Calibration Drift Under Reasoning (CDUR), hypothesis lock-in mechanisms, and dynamic token budgeting via the CABStop optimal stopping algorithm.

MSc thesis: evaluating & improving uncertainty calibration in retrieval-augmented LLMs. Compares 5 embedding models (BGE beats OpenAI) and cuts ECE ~32% via isotonic regression on real financial filings.

Calibration and uncertainty quantification for ranking systems.

Fine-tuning and calibrating the Chronos (T5) transformer for financial time series forecasting. Achieves up to 15% better accuracy (MASE) and reliable confidence estimates using perturbation-based consistency calibration (C3).

Research framework for analyzing calibration collapse under class imbalance in tabular machine learning, with adaptive calibration methods, stress-test benchmarks, and minority reliability evaluation.