Portfolio
17 projects with an active fraud AI launch stack: 5 synced repos and 7 live Cloud Run services
Modular enterprise AI platform comprising three focused systems: multi-agent orchestration (enterprise-agents), production infrastructure (enterprise-ai-infra), and hybrid RAG retrieval (enterprise-rag).
A secure federated learning framework that lets US states collaboratively train healthcare fraud detection models without sharing sensitive patient data.
Three-repo federated learning security ecosystem: core FL algorithms (federated-learning-core), adversarial attack/defense toolkit (fl-adversarial-security), and privacy-preserving ML components (privacy-preserving-ml).
Two-repo phishing detection platform: fedphish-platform for federated collaborative detection, and phishing-detection-engine for URL/domain/content-based ML classification.
LangGraph-based monitoring pipeline with 4 specialized fraud agents, deployed as API and dashboard services on Cloud Run.
Launch hub connecting 5 GitHub repositories to 7 live Cloud Run services, with deployment validation and service inventory tracking.
QLoRA fine-tuning pipeline for 3-class fraud narrative classification with a live Cloud Run interactive demo.
End-to-end fraud detection ML system combining feature engineering pipelines, multi-family model benchmarking (XGBoost, LightGBM, deep learning, anomaly detection), SHAP/LIME explainability, and production real-time scoring API.
Retrieval-augmented fraud investigation system with live API and UI deployment on Cloud Run, grounded responses, and source citations.
A static analysis engine with 95.7% accuracy that reverse-engineers and classifies zero-day banking trojans by extracting behavioral fingerprints.
Full-stack AI trading platform for the Indonesia Stock Exchange — FastAPI backend, 13-page Streamlit dashboard, IMSS multi-agent market simulator, ML prediction pipelines, and research workspaces with Playwright E2E coverage.
A production machine learning system that analyzes 227 million Medicaid records to automatically detect and flag complex healthcare fraud.
Production-style Medicaid fraud inference stack with Cloud Run API + dashboard deployment, batch scoring, and model health telemetry.
An interactive dashboard that makes $1.09 Trillion in healthcare spending transparent, allowing investigators to visually explore ML-flagged fraud anomalies.
This high-performance, statically generated portfolio website you are looking at right now, built from scratch with Astro, React, and Tailwind CSS.
Layered defense system combining ECDSA digital signatures, multi-factor anomaly detection, and time-decay reputation scoring to protect federated learning from Byzantine poisoning attacks.
Published steganography research implementing PVD+LSB+Modulo encoding with CTR_DRBG randomization and empirical image-quality evaluation.