Backend Systems / AI Infrastructure

Longchang Liu builds stable systems for agentic software.

Across past experiences at Alibaba (Amap), Tencent, and iMile, I focused on high-concurrency traffic governance and resilient service architecture. My work spans JVM internals, OS fundamentals, and production-grade AI workflows to make advanced capabilities predictable in real deployments.

Work Together View Archive Follow on X

AI Agents Deterministic and observable workflows

JVM / GC Runtime behavior and memory mechanics

Data Systems MySQL internals and query performance

Distributed Infra Fault-tolerant service design principles

// execution philosophy
const engineering = {
  reliability: "non-negotiable",
  latency: "measured, not guessed",
  architecture: "simple, then scalable",
  ownership: "from incident to root cause"
};

ship(engineering);

Featured Project

From idea to system-level implementation.

Lavis: macOS Multimodal AI Agent

A desktop AI agent that sees the screen, controls mouse and keyboard, and supports voice interaction. Designed as a closed-loop system with reflection and context compression for robust real-world execution.

Spring Boot React Electron TypeScript Java

Project Detail GitHub

Past Experience

Hands-on engineering across high-scale and AI-native environments.

Alibaba Group (Amap)

Oct 2025 - 2026

Backend Engineer Intern

Improving service resilience and traffic governance for map ecosystems, while architecting deterministic agentic workflows that can be trusted in production.

Tencent

Apr 2025 - Sep 2025

Data Engineer Intern

Optimized advertising data pipelines and built RAG-supported flows that grounded model outputs in verified performance signals.

iMile Logistics

Sep 2024 - Feb 2025

R&D Engineer Intern

Focused on order-system concurrency and MySQL tuning for cross-border logistics, balancing throughput with operational stability.

Core Capabilities

Systems thinking, measurable engineering, and practical AI architecture.

Java & JVMRuntime behavior, GC mechanisms, and performance diagnosis.

OS / xv6Memory, process isolation, and low-level execution fundamentals.

RAG EngineeringGrounded retrieval and controllable enterprise generation pipelines.

MySQL InternalsIndex, transaction, and query-path optimizations under load.

Distributed InfraConsistency, fault tolerance, and traffic-aware system design.

Agent WorkflowPlanning, acting, reflection, and guardrails for automation.

Let's Build Useful Systems

If you are building resilient backend infrastructure or real agentic products, I'd love to connect.

Email GitHub LinkedIn X / Twitter Archive