Building reliable, high-performance, and secure environments for Autonomous Agents.

I am an AI Platform & Infrastructure Engineer and Security Researcher specializing in autonomous agent runtimes, sandbox escape prevention, high-performance API gateways, and anti-detection automation.

Unlike traditional developers, I adopt a strictly AI-Native & Vibe Coding Workflow, leveraging agentic loops to accelerate engineering iteration while maintaining institutional-grade robustness in testing, sandboxing, and deployment.

• 🏗️ AI Infrastructure: Designing sandboxed runtimes (Node.js/Docker) and dual-SQLite messaging conduits.
• ⚡ Backend Engine: Creating concurrent, rate-limited FastAPI gateway routers and Redis distributed locking states.
• 🔒 Offensive Security & Evasion: Advanced binary program analysis, IDA Pro decompilation, and spoofing browser fingerprinters (TLS/Canvas/WebGL) to bypass cloud WAFs.

  Languages         : Python, Go, Node.js/TypeScript, C++, Bash
  AI Engine         : ReAct Loop, MCP (Model Context Protocol), Graph of Agents, Prompt Engineering
  Platform & Cloud  : Docker, Docker Compose, AWS (EC2, VPC, SG), Cloudflare (Workers, WAF, DNS)
  Data & Cache      : Redis (Lua locking, sliding rate-limiters), PostgreSQL, SQLite, SQLAlchemy 2.0
  Security Auditing : IDA Pro, Ghidra, Inline Hooking (Trampoline), Web fingerprint spoofing

I design systems by orchestrating agents and enforcing strict human-in-the-loop verification policies:

┌──────────────────────┐      ┌─────────────────────────┐      ┌─────────────────────────┐
│ Requirement Analysis │ ───> │ Task Decomposition Plan │ ───> │ Claude Code & Cursor    │
└──────────────────────┘      └─────────────────────────┘      └────────────┬────────────┘
                                                                            │
                                                                            v
┌──────────────────────┐      ┌─────────────────────────┐      ┌─────────────────────────┐
│ Production Deploy    │ <─── │ Multi-Agent E2E Testing │ <─── │ Human Approval Gate     │
│ CF WAF + AWS EC2     │      │ Pytest / Ruff Checks    │      │ (Permission Hooks)      │
└──────────────────────┘      └─────────────────────────┘      └─────────────────────────┘

Production-grade asynchronous AI ReAct loop gateway service.

• Concurrency Safeguard: Leverages Redis distributed locks via Lua scripts to stop race conditions on conversational states.
• SSE Stream & Connection Reclamation: Catches asyncio.CancelledError on client disconnection to instantly propagate termination signals up to LLM queries and release computing quotas.
• Observer RAG Evaluation: Computes real-time token intersection recall indexes dynamically in an asynchronous queue.

Hardened agent container hosting runtime based on Node.js and Docker SDK.

• Directory Traversal Fix: Implemented the isSafePath verification method in TypeScript MCP handler to stop arbitrary files from escaping out of /workspace/agent.
• Zero-Contention IPC: Dual-DB architecture (inbound.db / outbound.db) separating reader/writer processes to scale container tasks concurrently.

Anti-fingerprinting, high-entropy web scraping automation engine.

• Low-Level Spoofing: Rebuilt core Selenium capabilities using Camoufox browser to fake JA3/JA4 TLS handshake states, Canvas renders, and audio entropy.
• Bezier Trajectory Profiling: Custom interpolation simulating muscle deceleration to bypass Cloudflare behavioral anomaly WAF checks.

• Email: abyssrift1337@gmail.com
• GitHub: @Ruanyuxi1337