This is a sample code repository of the power transformer's health state (index) analysis or prediction by the regression model for experiment and learning purposes.

Sang-Wook's Development and Testing (SWDT)

A systems-thinking essay that explains why failure rarely happens suddenly. It shows how slow drift, accumulating pressure, and weakening buffers push systems toward collapse long before outcomes change, and why prediction-focused analytics miss the most important phase of failure.

A systems-thinking essay that reframes failure as a gradual transition rather than a discrete outcome. It explains how pressure accumulation, weakening buffers, and hidden instability precede visible collapse, and why prediction-based models arrive too late to prevent failure in human-centered systems.

Sang-Wook's Library (SWL)

Giving Every Modality a Voice in Microservice Failure Diagnosis via Multimodal Adaptive Optimization

[IJCNN'24] Official repository of "FashionFail: Addressing Failure Cases in Fashion Object Detection and Segmentation".

This repository hosts the project reports for the 'Applied FEA' course, which delves into the application of Finite Element Analysis (FEA) to a wide range of problems. It covers a variety of analyses, such as static, pseudo-static, dynamic, modal, buckling, contact, heat transfer, thermal stress, and thermal shock, all conducted using Ansys.

Hashin criteria predict failure modes in composites based on stresses in fiber and matrix. While effective, implementing them in numerical simulations poses challenges. Abaqus addresses this but is limited to 2D plane stress. A VUMAT subroutine extends Abaqus to simulate 3D damage progression using the energy method for static and dynamic problems.

Repo for "Large Language Model Reasoning Failures"

Automated AI-powered root cause analysis for GitHub Actions workflow failures.

FAILURE is a Matlab code to simulate the propagation of uncertainties in a plane stress problem subjected to uncertainties.

Public validation of SRI (Structural Retention) and CI (Collapse Index) metrics on AG News dataset: 90.4% accuracy, 9.2% flip rate, revealing model brittleness beyond standard benchmarks with AUC 0.874!

Public validation of Collapse Index (CI) on SST-2 dataset: 42.8% flip rate, AUC 0.698. Reveals model brittleness beyond 90%+ accuracy under perturbations!

In this project, a sandwich panel sample with a honeycomb structure under TNT blast load was analyzed. For this analysis, the Johnson-Cook elastoplastic model was applied, and the simulation was conducted under the load of a 1 kg TNT explosion.

Read more

Analysis of medical device failures and maintenance using Power BI and Python. Includes dashboards, failure trends, cost analysis, and data cleanup from SAP records.

Kubernetes operator for managing pod failure analysis in the Podmortem operator

GRAph Parallel Environment

Failure-mode research and constraint design for broken labor, hiring, and automated decision systems.