Research tools, exploitation frameworks, threat intelligence platforms, and CTF infrastructure.
Source code available upon request — reach out to discuss access.
Security and reliability engineering for an RFP intelligence pipeline (Python 3.11, FastAPI, n8n, pgvector): architected egress-guarded HTTP clients to mitigate SSRF, established a quarantined research LLM sandbox for injection resistance, and implemented a cross-model claim-verification layer. Hardened RAG reliability, unified error envelope patterns, directed dependency CVE management, and integrated gitleaks/bandit CI.
Resilient C2 infrastructure leveraging Tor hidden services for anonymised communications — domain fronting and traffic obfuscation to evade network detection. Privacy-hardened Linux with transparent Tor proxy and OverlayFS-on-tmpfs for zero forensic persistence via RAM-based wiping.
Adversarial testing on LLMs identifying 25+ vulnerabilities in content filtering and safety alignment. Developed novel jailbreaking techniques achieving 80% bypass rate against commercial guardrails. Documented prompt injection vectors for responsible disclosure to AI vendors.
ESP32-based vital signs monitor with MAX30102 (HR/SpO2) + DS18B20 (temp) acquisition and real-time streaming over WebSocket/REST on the LAN (mDNS). React (Vite) dashboard consumes a 4 Hz JSON stream, persists long-horizon history in IndexedDB (Dexie), and runs optional Gemini analysis via a local Express proxy to keep API keys off the client.
Led security architecture and contributed AI features for an AI career platform serving job seekers in Sub-Saharan Africa and MENA, submitted for IEEE TSYP 13 CS Challenge. Built the security layer from threat model (STRIDE) through implementation — encrypted-at-rest storage, stateless JWT sessions with 15-min/7-day rotation, CSP-enforced frontend, and compliance mapped to GDPR/NDPR/POPIA. Contributed to LLM pipeline design with prompt injection guards and PII-masking for safe resume processing.
Source code policy: All project source is available upon request for legitimate security research, hiring evaluation, or collaboration purposes.
Reach out at with a brief description of your context.