System Online

Intelligence over Noise.

Specialized research in AI Security, OT/ICS environments, and advanced Red Teaming operations. Designing robust architectures against emergent threats.

Research Initiative memory

ICSora Labs

An interactive OT security training and simulation platform designed to replicate industrial attack scenarios for hands-on defender training.

Focused on realistic ICS/OT attack paths, operational visibility, and practical security learning for industrial environments.
Built to help security teams transition from theoretical OT security knowledge into applied operational capability.

ICS/OT Simulation Security Training

View arrow_outward

Platform Engineering code_blocks

DVLA-VITA

A production-grade Capture The Flag platform simulating a government vehicle enquiry system to demonstrate real-world LLM vulnerabilities.

Engineered attack labs mapped to OWASP Top 10 for LLM Applications 2025 — Prompt Injection, Sensitive Information Disclosure, and Excessive Agency.
Implemented defensive controls: semantic filtering, PoLP API tokenization, input validation, and containerized prompt guardrails.

React · Vite · Python OWASP LLM 2025

View arrow_outward

Reference Material menu_book

ICS/OT Cheatsheets

Three quick-reference sheets distilling essential ICS/OT security knowledge for practitioners and defenders.

01 — Architecture & Network Fundamentals
02 — OT/ICS Reconnaissance & Penetration Testing
03 — Incident Detection & SOC Engineering

PDF Free Download

arrow_outward Sheet 01 — Architecture & Network arrow_outward Sheet 02 — Recon & Pen Testing arrow_outward Sheet 03 — Incident Detection & SOC

AI Engineering manage_search

Zscaler RAG Helper

Self-hosted RAG system for Zscaler incident response — engineers describe a problem in plain English and receive structured Root Cause Analysis, resolution steps, and source links grounded in 4,100+ official Zscaler help pages.

Hybrid dense+sparse retrieval (BAAI/bge-m3 + BM25) with cross-encoder re-ranking and zero-vendor-lock-in multi-provider LLM support (Groq, OpenAI, Anthropic, Ollama).
Local embeddings baked into Docker image (zero embedding API cost), single-command Compose deployment, and OpenAI-compatible API with ZIA/ZPA/ZDX product scoping.

Python · FastAPI · Docker LLM · RAG · Qdrant

View arrow_outward