تشخيص
SEASON ONE
CETHRAIAN-
X
2025 – PRESENT // BENCHMARK LEAD
Engineered an open, reproducible multi-label thoracic diagnostic framework using the NIH ChestX-ray14
repository and DenseNet-121 architectures. Trained across three independent seed initializations with
multi-metric verification: ROC-AUC, PR-AUC, expected calibration error (ECE), duplicate-patient leakage
audits, PA vs. AP projection subgroup analysis, and Grad-CAM++ saliency validation.
3 SEEDS
REPLICABILITY
14 LABELS
PATHOLOGY
0.00%
LEAKAGE RATE
SEASON TWO
تقييم
CETHRAIAN-
SHIFT
2026 – PRESENT // CROSS-DATASET EVALUATION
Built a frozen cross-dataset evaluation of three NIH-trained DenseNet-121 models on 15,000 VinDr-CXR
training images. Compared six thoracic findings under two radiologist-derived label policies while
preserving identical logits, thresholds, and calibration procedures across 10,722 Grad-CAM++ explanation
maps.
15,000
VINDR-CXR COHORT
10,722
GRAD-CAM++ MAPS
6 LABELS
EVAL POLICIES
نظام
SEASON THREE
ONYX
ROLEPLAY
2022 — 2023 // FOUNDER & PRINCIPAL SYSTEM ARCHITECT
Engineered, scaled, and managed an international multiplayer roleplay network accommodating 2,500–2,700
concurrent active users on the RAGE:MP multiplayer framework. Built custom distributed socket pipelines,
in-game economy databases, server synchronization, anti-cheat detection, and staff workflows.
2,700
PEAK CONCURRENT
99.9%
SYSTEM UPTIME
RAGE:MP
PLATFORM
SEASON FOUR
أمان
CYBERSECURITY
& APPSEC AUDIT
2020 — PRESENT // INDEPENDENT CYBERSECURITY & VULNERABILITY
RESEARCH
Conducted authorized application security audits, CTF security engagements, and OWASP top 10 vulnerability
assessments. Responsibly disclosed and verified remediation for 50+ web application and API security
vulnerabilities across production deployments.
50+
DISCLOSED & REMEDIATED
OWASP
TOP 10 AUDITING
100%
RESPONSIBLE DISCLOSURE
حوسبة
SEASON FIVE
SCIENTIFIC
COMPUTING
2020 — PRESENT // SCIENTIFIC COMPUTING & FULL-STACK
DEVELOPMENT
Development of reproducible research pipelines, automation tools, and web applications utilizing Python
(PyTorch, MONAI, Scikit-learn, OpenCV), TypeScript, Node.js, C#, and WebGL graphics engines. Engineered
for high reliability, clinical safety, and open science reproducibility.
PYTHON / C#
BACKEND RUNTIMES
PYTORCH
ML & MONAI
WEBGL
GRAPHICS ENGINES
