Available
AI CV Improver
Improve CV wording and role alignment while keeping it honest.
AI + Cybersecurity + Utility Platform
AI, Cybersecurity & Utility Tools Built to Solve Real Problems
Practical AI systems, security tools, productivity utilities, and technical projects designed for real-world use.
GTGTRepoZero Trust AISecurity UtilitiesResearch + Build
Platform Preview
Live DirectionAvailable
AI CV Improver
Beta
Phishing Analyser
Beta
Scam Detector
Beta
PDF Summariser
312
Security events
47
Tool modules
2,048
Checks today
Dashboard preview data is illustrative for interface direction.
// featured tools
AI & Security Tools
A focused set of practical utilities designed for everyday work, learning, and trust verification.
Beta
Phishing Email Analyser
Inspect suspicious emails for phishing indicators.
Beta
PDF Summariser
Summarise PDF content into key points.
Beta
AI Email Writer
Draft clearer professional emails from short prompts.
Beta
Scam Message Detector
Detect social engineering patterns in messages.
Available
Password Strength Checker
Assess password strength and security hygiene.
// tool categories
Explore by Category
AI Productivity Tools
Writing, summarisation, meeting and prompt support tools.
Cybersecurity Tools
Practical security checks for passwords, links, headers, privacy and risk.
Student Tools
Study, writing, citation and planning helpers for academic workflows.
File/PDF/Image Tools
Document and image utilities for conversion, OCR, and cleanup.
Money-Saving Tools
Cost comparison and financial utility helpers for better decisions.
AI Business Tools
Operational AI helpers for proposals, replies, invoices and policy drafts.
Verification & Trust Tools
Trust and authenticity checks for messages, reviews and sellers.
// about
A system builder, researcher, and evolving cybersecurity professional.
I am a London-based MSc Cybersecurity & Digital Forensics student originally from Bangladesh, focused on building practical cybersecurity systems and AI security infrastructure. GT is my personal technical identity, and GTRepo is where I document my research projects, experiments, and evolving understanding of modern cybersecurity systems.
My interests center around Zero Trust AI systems, explainable security decisions, digital forensics, threat monitoring, SOC-style platforms, and secure AI gateways. Moving from Bangladesh to London has made the work feel personal: I am building systems while learning, testing ideas through code, and staying curious about future technologies.
Cybersecurity Researcher
AI Security Builder
MSc Student — UWTSD London
Areas of Focus
Zero Trust Architecture
Designing and implementing Zero Trust principles for AI model serving — every request verified, no implicit trust.
AI Security & Gateways
Building security gateways that inspect prompts, evaluate risk, enforce policy, and audit AI model interactions.
Digital Forensics
Processing memory artefacts, normalising forensic evidence, and building investigation tools using Volatility3.
SOC & Threat Detection
Developing SOC-style dashboards for anomaly detection, attack-timeline visualisation, and threat correlation.
// research focus
What I'm Investigating
Zero Trust for AI model serving
Prompt & output security
Explainable policy decisions
Adaptive risk scoring
AI gateway & firewall design
Memory forensics & investigation
SOC monitoring & threat correlation
// currently learning
SOC Analyst PathEthical HackingNetworkingLinuxCloud DeploymentDockerNIST Zero TrustNIST CSFSecure AI Infrastructure
// research philosophy
My Philosophy
I am interested in how AI systems, infrastructure, human behavior, and security intersect. I believe future AI systems should be security-first by design: observable, explainable, monitored, and controlled through adaptive trust systems rather than blind access.
My work focuses on building practical systems that combine Zero Trust principles, behavioral analysis, explainable security decisions, forensic visibility, and secure AI infrastructure.
Security-first AI
AI systems should be designed with verification, policy, visibility, and control from the beginning.
Observable decisions
Security tools become more useful when their decisions can be inspected, explained, and challenged.
Adaptive trust
Access should respond to behavior, context, risk, and evidence instead of assuming anything is safe.
Forensic visibility
Good systems leave useful traces for investigation, learning, and future improvement.
// journey
Timeline
A grounded path from Bangladesh to London, through computer science, postgraduate cybersecurity, and practical AI security experimentation.
// projects
What I'm Building
Cybersecurity systems, forensic tools, and SOC infrastructure — all real builds.
Zero Trust AI Gateway
MSc Dissertation — Adaptive Zero Trust Architecture for Open-Source AI
Active
AI Security / Zero Trust
An adaptive Zero Trust security architecture for secure onboarding and usage of open-source AI models. Evaluates model posture before deployment, inspects every prompt for adversarial patterns, applies ALLOW / CHALLENGE / BLOCK policy decisions with explainable risk records, filters outputs, and continuously reassesses trust as sessions evolve — feeding a SOC-style monitoring dashboard.
Problem solved
Traditional perimeter-based security cannot address open-source AI risks: data leakage, adversarial manipulation, and misuse of models without a clear control point between user intent, policy enforcement, and audit visibility.
Architecture feel
Five-layer pipeline: model onboarding → Zero Trust enforcement → risk reduction (secure mode) → adaptive reassessment → explainability & SOC audit trail. Backend in FastAPI + PostgreSQL; frontend in Next.js.
Technologies
PythonFastAPIPostgreSQLSQLAlchemyReactTypeScriptNext.jsDocker
AI Firewall / API Interceptor
Experimental
Experimental
AI Gateway Security
An extension of the Zero Trust AI Gateway that lets external applications route model requests through a security layer before reaching their AI provider. Supports client API key authentication, prompt inspection, policy enforcement, output inspection, structured logging, and drop-in API-style integration.
Problem solved
External apps need a simple way to send AI traffic through security checks without redesigning their full application stack.
Architecture feel
Middleware-style API interceptor with client keys, request inspection, policy enforcement, provider routing, and structured event logging.
Technologies
FastAPIAPI GatewayClient API KeysAI FirewallDockerSecurity Middleware
MemScope Memory Forensics
Prototype / Research Build
Prototype
Digital Forensics
A digital forensics platform focused on ingesting and analyzing memory artefacts, correlating suspicious activity, and visualizing attack-chain investigations.
Problem solved
Memory evidence can be difficult to inspect quickly when process, network, DLL, command, and timeline signals are spread across raw artefacts.
Technologies
FastAPIReactPostgreSQLVolatility3
Repository coming soon
SOC Monitoring Dashboard
Active UI Module
Active
Security Operations
A SOC-style dashboard for visualising alerts, user anomalies, attack timelines, threat heatmaps, model posture events, and Zero Trust decision logs. Built as a standalone module that integrates with the AI Gateway backend.
Problem solved
Security decisions are hard to trust if they disappear into logs without a readable operational view.
Technologies
ReactTypeScriptThreat IntelligenceMonitoringDashboardSecurity Analytics
Personal AI Assistant — Aegis
Concept / Personal Lab
Concept
AI Assistant / Automation
A personal assistant concept focused on local-first automation, voice control, secure tool access, memory persistence, and privacy-aware AI orchestration. Designed to run privately without sending data to third-party services.
Problem solved
Personal assistants need useful automation without ignoring privacy, tool boundaries, local control, and security review.
Technologies
FastAPIReactVoice AssistantLocal AIAutomationSecurity
Repository coming soon
// platform purpose
Useful Tools Over Noisy Wrappers
Most online tools are bloated, low-quality, or overloaded with ads. GTRepo is being built as a modern AI, cybersecurity, and utility ecosystem focused on practical outputs: clear interfaces, trust-aware workflows, and tools that actually solve real problems.
// future ecosystem
Roadmap Direction
AI-assisted security tools
Practical AI pipelines that help identify suspicious behavior, risky input, and trust signals faster.
Developer APIs
API-first utility modules so other apps can plug into GTRepo checks and AI workflows.
Advanced productivity systems
Connected writing, summarisation, and task helpers built for real student and builder workflows.
Secure business automation
Automation tools with guardrails, audit visibility, and clear operational controls.
Browser extensions
Lightweight trust-check tools in-browser for scam, phishing, and authenticity workflows.
Intelligent utility systems
Cross-tool intelligence for verification, efficiency, and long-term platform learning loops.
// msc dissertation
My MSc dissertation explores a practical AI gateway architecture for secure model access. The system applies Zero Trust thinking to AI interactions by inspecting prompts, evaluating contextual risk, enforcing security decisions, inspecting outputs, and producing explainable logs that can support SOC-style monitoring.
AI securityZero TrustExplainable security decisionsBehavioral analysisSOC monitoringAI gateway architectureSecure model access
Research Summary
A student research system for testing how AI requests can be evaluated, explained, monitored, and controlled before model access is granted.
Screenshots / Placeholders
Dashboard and architecture visuals are kept as prototype previews until the dissertation build is ready to publish more fully.
Future Work
Improve behavioral scoring, expand threat intelligence signals, strengthen output inspection, and evaluate the gateway against realistic AI security scenarios.
// lab
GTRepo Security Lab
This lab contains my experimental cybersecurity builds, dissertation prototypes, SOC simulations, and AI security research interfaces. Systems shown here are research builds and student projects — not production deployments.
Research Prototype — metrics below are simulated demo data
sample
2,048
Demo Requests Evaluated
prototype
47
Policy Rules Loaded
active
4
Research Modules
demo
312
SOC Events (Simulated)
gtrepo-lab — research-prototypeRESEARCH BUILD
Simulated Lab Feed
DEMO
BLOCK14:23:01
High-risk prompt detected — policy engine decision: BLOCK
CHALLENGE14:22:47
Ambiguous request — escalated to human review queue
ALLOW14:22:31
Safe request forwarded to model provider — output checked
BLOCK14:21:55
Injection pattern matched in prompt — request rejected
INFO14:21:12
Audit trace logged — explainability record written to DB
Research Build Status
MSc Dissertation In Progress
Zero Trust Gateway
Building
MemScope Forensics
Prototype
AI Firewall
Experimental
SOC Dashboard
Active UI
Active experiments
- Adaptive prompt risk scoring
- ALLOW / CHALLENGE / BLOCK policy flows
- Output inspection and explainable audit records
Research prototypes
- Zero Trust AI Gateway
- AI Firewall / API Interceptor
- MemScope forensic correlation views
Deployment/testing notes
- Docker service composition
- Backend/API integration checks
- Dashboard states for simulated SOC events
Future ideas
- Behavioral threat intelligence
- Model posture monitoring
- More realistic security evaluation scenarios
// notes from the lab
Research Notes
Personal GT Lab notes on ideas, lessons, and technical reflections as GTRepo evolves.
Research notes2 min reflection
What is Zero Trust for AI?
Traditional Zero Trust focuses on users, devices, and infrastructure, but AI systems introduce new attack surfaces such as prompt injection, unsafe outputs, model misuse, and unrestricted access to powerful models. My current research explores how Zero Trust principles can be adapted into AI gateways that continuously inspect prompts, evaluate trust, monitor behavior, and apply explainable security decisions before and after model execution.
GT Lab Note
Research notes2 min reflection
Prompt Injection Thoughts
One of the most interesting AI security problems is prompt injection. Unlike traditional exploits, prompt injection manipulates instructions and model behavior rather than software memory directly. I am currently exploring how layered inspection, contextual risk analysis, policy enforcement, and behavioral monitoring can reduce these risks in AI systems.
GT Lab Note
Research notes2 min reflection
Docker Networking Lessons
While building and deploying projects, I realized many deployment issues are not caused by application code itself, but by networking and environment configuration. Understanding container networking, localhost isolation, environment variables, DNS resolution, and service communication became an important part of my learning journey.
GT Lab Note
Research notes2 min reflection
Building Security Dashboards
I enjoy designing security dashboards because they combine visibility, usability, and technical monitoring into a single interface. My recent work focuses on creating SOC-style dashboards that make threat activity, alerts, system posture, and security decisions easier to observe and understand.
GT Lab Note
Research notes2 min reflection
Explainable AI Security
Security systems become more useful when their decisions are understandable. I am particularly interested in explainable security decisions where systems can show why a request was blocked, challenged, or allowed instead of behaving like black boxes.
GT Lab Note
Research notes2 min reflection
Learning Threat Monitoring
My current learning path includes understanding how monitoring systems detect suspicious behavior over time. I am exploring how logs, alerts, behavioral patterns, and anomaly tracking can work together to improve visibility in modern security systems.
GT Lab Note
// beyond technology
Beyond Technology
Outside cybersecurity, I spend time exploring movies, anime, travelling, technology trends, and long-form curiosity around systems, infrastructure, and future technologies. I enjoy observing how technical systems, people, and ideas evolve together over time.
Movies and cinematic storytelling
Anime and imaginative worlds
Travelling and observing cities
Emerging technologies and AI systems
Long-form curiosity and experimentation
Systems thinking and future infrastructure
// stack
Technology Stack
The tools powering my cybersecurity builds, forensic systems, and AI security research.
Python
Core Language
FastAPI
Backend Framework
React
UI Framework
TypeScript
Type-Safe Frontend
PostgreSQL
Database
Docker
Containerisation
SQLAlchemy / Alembic
ORM & Migrations
Hugging Face
AI Models
NIST Zero Trust
Security Framework
Volatility3
Memory Forensics
Linux
Operating System
Vercel / Render
Deployment
GitHub
Version Control
// contact
Get in Touch with Shihab
Whether you're a recruiter, a supervisor evaluating my dissertation, a collaborator on security tools, or just curious about the work — I'm open to conversations.
London, United Kingdom
// for visitors
Why Are You Here?
For Recruiters
- SOC analyst roles
- Cybersecurity graduate positions
- AI security engineering
- Digital forensics
- Security research
For Supervisors & Academics
- MSc dissertation — Zero Trust AI Gateway
- Research evaluation and feedback
- Methodology discussions
- NIST Zero Trust alignment
- Collaboration on AI security research
For Collaborators
- AI security tooling
- Web security systems
- Forensic investigation tools
- SOC dashboard development