projects

Selected engineering and research work.

engineering

ChangeSafe: An AI Infrastructure Change Airlock

A v0.1 safety boundary that converts AI-generated infrastructure recommendations into typed, schema-validated, transactionally applied patches with deterministic policy checks and hash-bound decision receipts.

research
project thumbnail

Epistemic Parity: Reproducibility as an Evaluation Metric for Differential Privacy

A benchmark and evaluation for reproducibility in differential privacy with state-of-the-art DP synthesizers.
project thumbnail

Fairness-Enhancing Interventions in ML Pipelines

Exploring fairness-enhancing methods in binary classification pipelines using AIF360 and Folktables datasets.
project thumbnail

Out-of-distribution Robustness Evaluation of State-of-the-art Vision Models

Research project comparing the robustness of 58 computer vision models
project thumbnail

Explaining Predictions with SHAP

Using SHAP to interpret predictions from a classifier trained on the 20 Newsgroups dataset.
project thumbnail

Nutritional Labels for Automated Decision Systems by Home Credit Default Risk

A fairness and explainability analysis on loan repayment predictions using machine learning models.
project thumbnail

Privacy-Preserving Synthetic Data

Generating privacy-preserving synthetic datasets and evaluating their accuracy and fairness using Data Synthesizer and MST.
project thumbnail

Spiking Neural Network Simulatation: Modeling and Synchronization Analysis

A Python-based simulator exploring spiking neural networks and synchronization properties in neuronal systems.
project thumbnail

Time-series Medical Image Classification

Image Classification Model for Temporal Disease Progression of Chest X-ray dataset