AI Security + Post-Quantum Cryptography. OWASP LLM Top 10 | MITRE ATLAS | NIST PQC (ML-KEM, ML-DSA) | Privacy-First RAG | Rutgers University

I build secure AI systems for regulated research workflows, with emphasis on privacy-first RAG, PHI/PII de-identification, audit evidence, fail-closed data controls, secure retrieval, and human-in-the-loop AI governance.

• Secure AI systems for regulated research workflows
• Privacy-first RAG and secure retrieval pipelines
• PHI/PII de-identification and synthetic privacy testing
• LLM/RAG security, audit evidence, and human-in-the-loop controls
• AI-assisted secure SDLC and threat modeling
• Long-term research direction: post-quantum cryptography readiness and crypto-agility

A privacy-first AI research workflow suite for clinical research environments.

1. RePORT-AI-Portal — privacy-first local RAG assistant for PHI-scrubbed clinical research bundles.
2. RePORT-agent — LangGraph multi-agent workflow for secure research analysis and human review checkpoints.
3. PHI-Handling-IRB-Review-Support — synthetic PHI/PII corpus, validation harness, and IRB review support tooling.
4. MetaScope — research software contribution supporting scientific workflow development and manuscript preparation.
5. Secure-AI-Flow — security-first methodology for AI-assisted software delivery, governance, and release evidence.

Languages: Python, SQL, R, Bash
AI / LLM: RAG, LangGraph, Hugging Face Transformers, local-first AI workflows
Security / Privacy: PHI de-identification, PII detection, audit logging, encrypted mapping storage, fail-closed controls, secure retrieval
Infrastructure: Docker, Singularity, GitHub Actions, Linux, HPC environments, Nextflow
Frameworks: HIPAA Safe Harbor-informed handling, India DPDPA-informed handling, NIST AI RMF concepts, OWASP Top 10, secure SDLC

• Public repositories do not contain real PHI or patient-identifying data.
• PHI/PII examples are synthetic or sanitized.
• Privacy and compliance language is implementation-oriented and review-support focused, not a legal certification.
• IRB approval has not been received for the separate PHI-handling review-support system.

Secure AI systems, LLM/RAG security, privacy-preserving AI workflows, cybersecurity automation, AI governance evidence, post-quantum cryptography readiness, crypto-agility, and quantum-resilient security architecture.

AI Security Engineer, Secure AI Systems Engineer, Product Security Engineer, Security Automation Engineer, Privacy Engineering, and Research Software Engineering roles involving secure AI or regulated data.

LinkedIn: https://www.linkedin.com/in/solomon-s-joseph/
GitHub: https://github.com/solomonsjoseph