B.Sc. Engineering Thesis • Wrocław University of Science and Technology (2025 – 2026)
Adaptive Moving Target Defense (MTD) in Containerized Environments
B.SC. THESIS PROJECT
Designed and implemented an autonomous, SIEM-driven Moving Target Defense testbed that eliminates static infrastructure predictability. By closing the OODA loop between Splunk Enterprise telemetry and a host-isolated Python/Docker controller, the system detects application attacks (SQLi, RCE, Brute-Force), seamlessly migrates legitimate traffic across twin stateless containers, and wipes compromised OverlayFS layers without dropping active user sessions.
277 ms
Avg Nginx SIGHUP Proxy Switch
0 / 32,414
HTTP 502/504 Gateway Errors
100.0%
Redis Session Continuity (300 migrations)
38.8 s
Mean Attacker Dwell Time (100% wipe)
Key Architectural Highlights
- Closed-Loop OODA Automation: Custom SPL correlation rules in Splunk Enterprise detect SQLi, Command Injection, and HTTP 401 Brute-Force anomalies, triggering JSON webhooks to the MTD controller.
- Zero-Downtime Traffic Shifting: Active-Passive upstream switching via Nginx graceful worker reload (
SIGHUP) paired with a 3-second grace period before SIGKILL + --force-recreate container reincarnation.
- Stateless Compute & Session Persistence: Followed Twelve-Factor App principles by decoupling session UUIDs into external Redis (
SETEX) and persistent records into PostgreSQL 15, ensuring upperdir wipes destroy attacker webshells/backdoors while preserving user logins.
- Control Plane Isolation: Isolated
mtd_engine.py and /var/run/docker.sock on the host OS outside container namespaces, mounting Nginx logs read-only (:ro) into Splunk to prevent log tampering.
- Custom Splunk SOC Dashboard: Built a multi-tab Simple XML operational dashboard tracking switch latency (ms), reactive vs. proactive rotations, attack payloads, and node load distribution.
System Topology (mtd_bridge_net)
Click to enlarge
Python 3.11
Docker & Compose
Docker SDK
Splunk Enterprise (SPL)
Nginx Reverse Proxy
Flask
Redis
PostgreSQL 15
OverlayFS
Bash
Desktop & Backend System Architecture • 2024
PDW — Secure Records Management System
DOCKERIZED STACK
Multi-container desktop records management system modeled after law enforcement databases for tracking individuals, vehicles, criminal cases, and court sentences. Combines a native PyQt6 graphical client with a layered FastAPI REST backend and relational PostgreSQL 17 storage orchestrated via Docker Compose.
Core Engineering & Security Features
- Layered FastAPI Backend: Clean separation across routing (
main.py), Pydantic schema validation (e.g., strict 11-digit PESEL validation), SQLAlchemy ORM models, and CRUD business logic with auto-generated Swagger/OpenAPI docs.
- Authentication & Audit Trail: Multi-factor credential verification (
login, password, badge_number) backed by bcrypt password hashing, HS256 JWT session tokens, and comprehensive action logging in a dedicated logs audit table.
- Normalized Relational Schema: 9 interconnected PostgreSQL 17 tables (
person, appearance, vehicle, criminal_file, sentence, crime_database, health_and_operations, user, logs) enforced by foreign key constraints.
- PyQt6 Desktop Client & Containers: Multi-screen GUI (
QStackedWidget) supporting multi-criteria filtering, photo rendering via Pillow/ImageQt, CSV bulk imports, and single-command Linux X11 deployment via run.sh.
Container Architecture
Click to enlarge
Python 3.11
FastAPI
PyQt6
PostgreSQL 17
SQLAlchemy
Pydantic
JWT & bcrypt
Docker Compose
Adminer