| Project | What it proves |
|---|---|
 |
Production-grade REST backend β JWT rotation with reuse detection, RBAC, rate limiting, pytest suite, Docker Compose |
 |
One OpenAI-compatible endpoint over GLM + Qwen + DeepSeek + custom APIs β load balancing, failover, streaming |
 |
Full-stack product: Node.js API + realtime WebSockets + Vue 3 SPA + SQLite + Docker, bilingual RTL |
 |
Cryptographic engineering: X25519 handshake, ChaCha20-Poly1305, forward secrecy β auditable pure Python |
 |
Postman mastery: chained CRUD contracts, HMAC request signing, data-driven runs, Newman CI reports |
 |
ML meets production: transformer backend + deterministic fallback, batch inference, Prometheus metrics |
Also worth a look: BaleForge (async bot framework) Β· Snaplink (Next.js + Prisma) Β· HabitGrid (Expo app) Β· BalePhp (PHP/Laravel SDK) Β· NetPilot (single-binary Go network toolkit) Β· ModbusLite (Modbus TCP for .NET 8)
I'm a generalist who dives deep. My work sits at the intersection of modern web engineering, AI infrastructure and industrial automation. I architect cross-platform systems, build NLP/GenAI services, program PLCs β and I approach every problem with a security researcher's mindset: threat-model first, then build.
I ship in public: every repository linked above is a working build with documentation, tests and deployment scripts β no vaporware.
-F59E0B?style=flat-square&logo=codeforces&logoColor=white){kind=icon}
π¨ Frontend
|
π± Mobile & Desktop
|
βοΈ Backend
|
ποΈ Database
|
π§ AI / ML
|
π§° Tools
|
Security isn't a checkbox here β it's the workflow: every repo runs secret scanning + push protection, dependencies are covered by Dependabot automated fixes, credentials are env-only, and sensitive flows (JWT rotation, RBAC, rate limiting) are implemented and tested in the open. Security-audit tooling (gitleaks, custom pattern scanners) is part of my routine.
