I'm a DevOps/SRE engineer focused on building high-performance distributed systems.

Infrastructure: Kubernetes, Istio, Docker, Terraform
Monitoring: Prometheus, Grafana, Thanos, ClickHouse
Cloud: AWS
Languages: Rust, Go, TypeScript, Python

A custom Layer 1 blockchain with Proof-of-Contribution consensus, DeFi, NFT marketplace, and full wallet ecosystem.

graph LR
    Wallet[Wallet] -->|RPC| Node[QFC Node]
    Explorer[Explorer] -->|API| Node
    Node -->|P2P libp2p| Node2[QFC Node]
    Node --> Storage[(RocksDB)]
    DeFi[DEX / DeFi] -->|Contracts| Node
    NFT[NFT Marketplace] -->|Contracts| Node

Core: qfc-core — Rust blockchain node with Blake3 PoW, Merkle Patricia Trie, Ed25519/VRF cryptography, libp2p networking
Ecosystem: Explorer, DEX, NFT marketplace, wallet (desktop + mobile), faucet, AI inference router, SDK (JS/Python), VS Code extension
34 repositories covering the full blockchain stack

Tech stack: Rust, RocksDB, libp2p, Blake3, Ed25519, React, TypeScript, Solidity

Lightweight VPS fleet management platform — track instances across dozens of cloud providers, manage IP addresses, and monitor with Prometheus/Grafana.

graph LR
    Web[React UI] -->|API| Server[Rust + Axum]
    Server --> PG[(PostgreSQL)]
    Agent1[sigma-agent<br/>VPS-1] -->|heartbeat| Server
    Agent2[sigma-agent<br/>VPS-N] -->|heartbeat| Server
    Prom[Prometheus] -.->|scrape| Agent1
    Prom -.->|scrape| Agent2

Components: sigma-api (Rust/Axum), sigma-web (React), sigma-agent, sigma-probe, sigma-cli
Tech stack: Rust, Axum, PostgreSQL, Redis, Prometheus, Grafana, Thanos, React, Docker, K8s

A self-built content delivery network with edge caching, intelligent scheduling, and security.

graph LR
    User --> Scheduler[DNS Scheduler]
    Scheduler --> Edge[Edge Node]
    Edge --> Origin[Origin]
    Console[Admin UI] -->|REST API| Control[Control Plane]
    Control -->|gRPC| Edge

Highlights:

• Edge Node (cdn-edge) — 32K QPS caching proxy with two-tier LRU cache, WAF, HTTP/3, Gzip/Brotli, WebSocket, image optimization
• Control Plane (cdn-control) — REST API + gRPC config sync, JWT + TOTP 2FA, audit logging
• Scheduler (cdn-scheduler) — DNS + HTTP 302 routing with GeoIP, health checks, load-weighted selection
• Admin Console (cdn-console) — React + TypeScript + Ant Design dashboard
• 8 repositories covering edge proxy, control plane, scheduler, console, CLI, deployment, Terraform, design docs

Tech stack: Go, gRPC, PostgreSQL, Redis, ClickHouse, Prometheus, React, Docker, K8s

A bare-metal operating system purpose-built for LLM inference — zero overhead, maximum throughput.

graph LR
    App[LLM Inference] --> Kernel[merlion-kernel]
    Kernel --> GPU[GPU Direct Access]
    Kernel --> Mem[Memory Manager]
    Go[Go Programs] -->|go-merlionos| Kernel
    Rust[Rust Programs] -->|libmerlion| Kernel
    C[C/C++ Programs] -->|musl-merlionos| Kernel

Core: merlion-kernel — Custom kernel with GPU-first scheduling, zero-copy memory management
Language support: Go (go-merlionos), Rust (libmerlion), C/C++ (musl-merlionos)
Inference: merlion-infer — Bare-metal LLM inference engine
Zig port: merlionos-zig — Clean reimplementation in Zig
Try it: Playground

Tech stack: Rust, Zig, Assembly, C, Go, GPU/CUDA