DDevineSecurity Engineer

About

I learn security by building, testing, and breaking things.

Open to roles — Nairobi / Remote52 tests · P3 hardened · 0 CVEs

I'm Devine Nyaenya, a self-taught security builder based in Nairobi, Kenya. I build security systems that don't just look secure — they prove it.

My path has been deliberately hands-on. Through Cisco's cybersecurity curriculum and a long string of practical projects, I've focused on the problems that make systems safe in practice: how you authenticate and authorize actors, how you enforce least privilege and separation of duties, and how you keep an audit trail that cannot be silently rewritten.

My strongest project, Chokepoint, is a least-privilege access-control and tamper-evident audit platform for high-impact operations — for both humans and AI agents. It grew out of an earlier security simulation lab where I first proved these controls by attacking my own design. 52 tests, 6/6 attacks detected, P3 hardened with Argon2id, HMAC, TOTP, SIEM shipping.

I care especially about the frontier of securing autonomous AI agents: as agents start acting with borrowed human authority, identity and privilege abuse becomes one of the most important security problems in software. Chokepoint is built around exactly that question — OWASP Agentic AI ASI03.

When a project is incomplete, I say so. I'd rather show an honest work-in-progress with real lessons than a polished overclaim — in security, overstating a control is itself a vulnerability.

52
tests
json + sqlite
6/6
detection
9 precise alerts
P3
hardened
Argon2 HMAC TOTP SIEM

Methodology

Build → Test → Break → Learn → Secure

My loop: implement controls, write tests that attack them, fix what breaks, document honest limitations, ship proof alongside code.

1

Build

Implement RBAC, four-eyes, hash chain, HMAC, TOTP

2

Test

52 tests: workflow, detection, negative, P3

3

Break

Attacker sim: 6 attacks + ledger tamper + self-approval

4

Learn

14 honest limitations, false positives 14→9

5

Secure

Ship with CI, demos, threat model, release