I build AI systems and production software, with a cybersecurity degree underneath that shapes how I design them — threat models before features, auditable decisions over black boxes.

Most of my work sits where those two things overlap: models that have to explain themselves, APIs that have to survive abuse, pipelines that have to run on real traffic. My most-starred project, the AI Vulnerability Scanner, runs that whole loop end to end — recon → semantic CVE matching → generated remediation.

• BS Cybersecurity, FAST NUCES Islamabad — Class of 2026
• AWS Certified AI Practitioner, plus 5 AWS Academy graduate badges across cloud, security, and CI/CD
• Currently deep in applied ML — transformer architectures, RAG pipelines, and model explainability (SHAP/LIME)
• Islamabad, Pakistan · open to AI/ML engineering and software engineering roles
• Reach me at areejzeb04@gmail.com

Project	What it does	Stack
AI Vulnerability Scanner	Scans a host with Nmap, matches each discovered service to relevant CVEs using SBERT embeddings, classifies severity, then generates tailored remediation steps with a fine-tuned T5 model.	Python FastAPI SBERT T5 Nmap
Aegis — Threat Detection & XAI	Intrusion detection over streaming network traffic using CNN-LSTM ensembles, with a RAG pipeline that enriches alerts into analyst-ready summaries and a SHAP-based XAI inspector so every detection is explainable.	Python PyTorch RAG SHAP React Docker
Distributed DDoS Detection	Parallel network-flow anomaly detection across MPI nodes with optional CUDA acceleration — entropy, PCA and CUSUM algorithms feeding real-time blocklisting and BGP FlowSpec/ACL rule generation. 10K+ flows/sec at ~95% accuracy.	C MPI CUDA PCA/CUSUM
Secure University Chatbot	Role-based FAQ assistant built as a secure-SDLC reference: Argon2id hashing, server-side sessions, RBAC, hash-chained audit logs, rate limiting, and TLS termination via nginx.	FastAPI Streamlit nginx RBAC
Filmception	Multilingual film-genre classifier: BERT over plot summaries, plus a translation + TTS layer that narrates summaries in Urdu, Arabic, and Korean.	Python BERT NLP TTS
Image Moderation API	Content-safety service for user uploads — token-authenticated moderation endpoints, ML safety checks, request logging, and a minimal review frontend.	FastAPI MongoDB Sightengine

• Pharmacy ERP & POS — full-scale ERP with dual POS/back-office interfaces: inventory, configurable EOD financial reporting, MongoDB transactions, RBAC, GS1 DataMatrix barcode parsing, ESC/POS thermal printing, and n8n workflows for expiry monitoring, supplier demand, and audit logging. Next.js 14 TypeScript MongoDB Docker Vitest n8n
• Real-Time ASL Recognition — sign-language recognition with Temporal Transformers over a MediaPipe 3D keypoint pipeline, deployed as an edge-optimized ONNX model hitting 30 FPS at 80%+ accuracy fully client-side. PyTorch MediaPipe ONNX

AI / ML

Languages

Backend & Frontend

Cloud & DevOps

Security

Also completed via AWS Skill Builder: AWS Security Fundamentals (2nd Ed.), Shared Responsibility Model, Security Groups vs. NACLs, Protecting Your Instance with Security Groups.

See all verified badges on Credly →

Cybersecurity Intern · ExplorerBees, Islamabad · Jul – Oct 2025 Ran a full penetration-testing audit of authentication APIs, simulating brute-force and OTP-flooding attacks to validate backend resilience. Documented high-risk findings and delivered a remediation roadmap aligned to the OWASP API Security Top 10.

IT Intern · Haidri Beverages (PepsiCo Pakistan), Islamabad · Jul – Aug 2025 Reviewed and validated ISO 27001 audit findings across an enterprise SAP environment, tracked remediation progress, and authored a risk-assessment report. Built a custom ABAP program automating HR data management workflows.

IT Intern · ConnectX, Islamabad · Jul – Aug 2024 Designed a high-fidelity website prototype in Figma with user-centric layouts and interactive components to improve digital branding.