DevSecOps & Cybersecurity Engineer — Python · ML Security · Secure CI/CD
I build tooling that detects and blocks attacks against LLM applications. Software engineer based in Cameroon, working at the intersection of application security and machine learning.
Open to relocation to Canada and to remote-first roles.
mcp-scrutiny — a security scanner for Model Context Protocol servers. Finds unpinned packages, rug pulls, tool poisoning, cross-server shadowing and toxic flows. Detection is semantic, not keyword-based: a trained 31-feature model scores each tool description and reports which features drove the decision, with a decode-and-rescore pass for encoded payloads. Fully offline, zero third-party dependencies, SARIF output for CI. Published October 2026.
llm-sentinel — an AI security gateway. A fine-tuned lightweight model that flags prompt-injection attempts before they reach the target model. PyTorch, FastAPI, Docker, full CI pipeline and test suite.
cmg-exploitation — production and maintenance tracking for an industrial group operating four sites. Python, PostgreSQL/Supabase, role-based access.
- Application & AI security — threat modelling, prompt-injection and LLM attack surfaces, secure pipeline design
- DevSecOps — CI/CD hardening (GitHub Actions), Docker, Linux, secrets and access management
- Software engineering — Python, FastAPI, clean architecture, object-oriented design, automated testing
- Machine learning — PyTorch, model fine-tuning, neural network architectures, applied AI
- Data — SQL, PostgreSQL, ETL pipelines, business intelligence
Because I come from an engineering background, I don't stop at a notebook: I ship modular, secure, production-ready systems.
- Building and hardening mcp-scrutiny, my MCP security scanner
- Building and hardening llm-sentinel, my LLM security gateway
- Implementing foundational deep learning papers from scratch in PyTorch
- Looking to collaborate on open-source ML security and evaluation tooling
- Learning: AI red-teaming, LLM fine-tuning, distributed training
Ask me about securing ML pipelines, DevSecOps practice, and the security of LLM applications.
- Email: bryandtamouffe@gmail.com
- Open to: DevSecOps, application security and ML engineering roles — Canada, Europe, or remote
"Quality is not an act, it is a habit."
