Platform Development & System Software Engineer平台开发与系统软件工程师

Liu Tianyi

Focused on system software and engineering infrastructure abstracting complex, repetitive system-level work like driver builds, firmware tests, crash analysis, and performance tuning into reusable, traceable, scalable automation tool-chains专注于系统软件与工程基础设施建设 擅长将驱动构建、固件测试、崩溃分析、性能调优等复杂且重复的系统级工作 抽象为可复用、可追踪、可规模化运行的自动化工具链

Drawn to the engineering patterns behind the data continually exploring how AI Agents and knowledge engineering land in system diagnostics freeing teams from heavy manual analysis to locate issues faster, ship software more reliably, and drive innovation more efficiently关注数据背后的工程规律 也持续探索 AI Agent 与知识工程在系统诊断中的落地 让团队从繁重的人工分析中解放出来 更快定位问题、更稳定交付软件、更高效地推动创新落地

View Projects查看项目 Get In Touch联系我

About关于

A bit about me关于我

I'm Liu Tianyi — a system-software and engineering-infrastructure engineer based in Shanghai.我是刘天义，一名常驻上海的系统软件与工程基础设施工程师。

My work has long centered on drivers, firmware, test infrastructure, and system-level problem diagnosis. At Lenovo I own the drivers and platform tooling around the Windows core stack — Display, Graphics, and Power — backing validation, delivery, and root-causing across the Yoga and ThinkBook lines. Before that, at Quanta, I built the EFI / UEFI firmware test infrastructure for Apple MacBook and worked directly with Apple's engineering team on firmware and hardware-adaptation issues.我的工作长期围绕驱动、固件、测试基础设施与系统级问题诊断展开。在联想，我负责 Windows Display、Graphics、Power 等核心栈相关的驱动与平台工具，支撑 Yoga 与 ThinkBook 产品线的验证、交付和问题定位；此前在广达，我为 Apple MacBook 搭建 EFI / UEFI 固件测试基础设施，并直接与 Apple 工程团队协作解决固件与硬件适配问题。

I like breaking complex problems into clear engineering models, and I habitually turn slow, repetitive, experience-dependent processes into automated systems. Driver builds, firmware tests, WHQL certification, Windows Update release, BSOD crash analysis, trace-data processing, performance and power tuning — this seemingly scattered work can all be abstracted into stable, reusable, traceable engineering infrastructure.我喜欢把复杂问题拆成清晰的工程模型，也习惯把缓慢、重复、依赖人工经验的流程改造成自动化系统。驱动构建、固件测试、WHQL 认证、Windows Update 发布、BSOD 崩溃分析、Trace 数据处理、性能功耗调优——这些看似分散的工作，本质上都可以被抽象为稳定、可复用、可追踪的工程基础设施。

I move comfortably across driver validation, Windows services, automated testing, system diagnostics, performance analysis, and cross-vendor collaboration; I also enjoy using full-stack web, Docker, CI/CD, and all kinds of internal tools to turn ideas into working systems fast. C#, C++, Python, Go, and Node.js are my everyday languages — and AI Agents, knowledge engineering, and tool-calling are becoming my new way to scale engineering efficiency.我可以在驱动验证、Windows 服务、自动化测试、系统诊断、性能分析与跨厂商协作之间切换，也乐于用全栈 Web、Docker、CI/CD 和各种内部工具，把想法快速落成可用的系统。C#、C++、Python、Go 与 Node.js 是我常用的语言，而 AI Agent、知识工程和工具调用，则正在成为我扩展工程效率的新方式。

I believe great engineering systems don't just get the job done — they cut repetitive labor, accumulate expertise, and help a team understand problems, validate hypotheses, and deliver results faster.我相信优秀的工程系统不只是完成任务，更应该减少重复劳动、沉淀专业经验，并让团队更快地理解问题、验证假设和交付结果。

Work作品

Projects项目

Platforms and engineering infrastructure I've built — from automated crash diagnosis and driver delivery pipelines to trace-based analysis tools and internal systems that turn manual engineering work into repeatable, scalable workflows.我构建过的平台与工程基础设施——从自动化崩溃诊断、驱动交付流水线，到基于 Trace 的分析工具与内部工程系统，将依赖人工经验的流程沉淀为可复用、可规模化运行的工作流。

Skills技能

Skillset技能栈

Core systems, debugging, automation, and infrastructure tools I use to build reliable engineering workflows.我用于构建可靠工程工作流的系统软件、调试、自动化与基础设施技术栈。

C / C++ Systems-level tooling, driver-adjacent infrastructure, performance-sensitive utilities, and Windows core-stack work.系统级工具、驱动相关基础设施、性能敏感型工具，以及 Windows 核心栈开发。
Python Automation, trace/log processing, test tooling, and the LLM + WinDbg analysis engine behind diagnostic platforms.自动化、Trace/日志处理、测试工具，以及诊断平台背后的 LLM + WinDbg 分析引擎。
C# / .NET Internal platforms, services, and desktop tooling.内部平台、服务与桌面工具。
Go Lightweight backend services, internal APIs, and deployable CLI utilities for engineering workflows.轻量后端服务、内部 API，以及面向工程工作流的可部署 CLI 工具。
AI Agents Skills + Loop task-driven agents, knowledge retrieval, tool-calling workflows, and AI-assisted root-cause analysis.Skills + Loop 任务驱动 Agent、知识检索、工具调用工作流，以及 AI 辅助的根因分析。
CI/CD Build, test, and release pipelines for drivers, firmware, and platforms.驱动、固件与平台的构建、测试、发布流水线。
WinDbg / ETW / ETL Driver-stack root-cause analysis: crash dump inspection, symbol-level debugging, and trace-based diagnostics.驱动栈根因分析：崩溃转储检查、符号级调试，以及基于 Trace 的诊断。
Perf / Power PowerLimit tuning, benchmark analysis, performance-regression investigation, and MSR/MMIO-based platform counters.PowerLimit 调优、Benchmark 分析、性能回退排查，以及基于 MSR/MMIO 的平台级 Counter。
WHQL / HLK Driver certification, multi-arch testing, signing, Windows Update.驱动认证、多架构测试、签名与 Windows Update。
Docker / Hyper-V Reproducible build, test, and lab environments.可复现的构建、测试与实验环境。

Experience经历

Where I've worked工作经历

Lenovo

Nov 2022 — Present

Software Engineer — Test Infrastructure / System Software软件工程师 — 测试基础设施 / 系统软件

  • Core engineer for Windows platform and driver infrastructure across Lenovo consumer products — covering Display, Graphics, and Power modules for Yoga and ThinkBook validation, delivery, and mass-production support.联想消费级产品线 Windows 平台与驱动基础设施核心工程师，覆盖 Yoga 与 ThinkBook 的 Display、Graphics、Power 模块验证、交付与量产支持。
  • Drive system performance and power tuning across these platforms, including dynamic PowerLimit strategies for real-world scenarios, benchmark and competitive analysis, and stutter/jank root-cause investigation using ETW/ETL traces and MSR/MMIO-based performance counters.负责平台级性能与功耗调优，包括面向真实使用场景的动态 PowerLimit 策略、Benchmark 与竞品分析，以及基于 ETW/ETL Trace 和 MSR/MMIO Performance Counter 的卡顿 / 性能异常根因定位。
  • Designed and shipped CI/CD and automated-test infrastructure covering driver build, validation, certification, and release, reducing multi-hour manual delivery flows to minutes.设计并落地覆盖驱动构建、验证、认证与发布的 CI/CD 和自动化测试基础设施，将原本数小时的人工交付流程压缩至分钟级。
  • Own WHQL certification and release for 10+ drivers: HLK setup, multi-arch loop testing, signing, and Windows Update — 100% compliant delivery.全面负责 10+ 驱动的 WHQL 认证与发布：HLK 搭建、多架构循环测试、签名与 Windows Update——100% 合规交付。
  • Collaborate closely with Intel and Microsoft engineering teams to root-cause driver, OS, platform, and compatibility issues, and to deliver system-level features that improve reliability, performance, and end-user experience.与 Intel、Microsoft 工程团队紧密协作，联合定位驱动、OS、平台与兼容性问题，并推动提升可靠性、性能和终端体验的系统级功能落地。
  • Built a 0-to-1 BSOD intelligent-diagnosis platform using a Skills + Loop task-driven agent architecture, integrating dump parsing, symbol resolution, issue data, and knowledge retrieval to infer and classify crash root causes.从 0 到 1 构建基于 Skills + Loop 任务驱动架构的 BSOD 智能诊断平台，整合 Dump 解析、Symbol 还原、issue 数据与知识检索，实现崩溃根因推断与自动分类。
  • Reduced average localization time by 80%, bringing AI-assisted debugging into test and analysis workflows and shifting the team from manual test execution toward platform-driven engineering.将平均定位周期缩短 80%，把 AI 辅助调试引入测试与分析流程，推动团队从人工执行测试转向平台化工程运营。

Quanta

Oct 2020 — Feb 2022

Firmware Test Infrastructure Engineer — EFI / UEFI & System Debugging固件测试基础设施工程师 — EFI / UEFI 与系统调试

  • Owned EFI/UEFI firmware test infrastructure and QA for Apple MacBook — cross-site, cross-region test flows and multiple Dry Run cycles before mass production.负责 Apple MacBook 的 EFI/UEFI 固件测试基础设施与质量保障——跨站点、跨地域测试流程，量产前多轮 Dry Run。
  • Root-caused complex firmware and system issues from logs and core dumps; built layered validation and regression plans to speed R&D fixes.基于日志与 Core Dump 对复杂固件与系统问题做根因定位；制定分层验证与回归方案，加速研发修复。
  • Served as the EFI test technical interface across hardware, BIOS, and key-component teams, working directly with Apple engineering on version iteration, issue triage, and quality evaluation.担任 EFI 测试技术接口人，协同硬件、BIOS 与关键部件团队，并直接对接 Apple 工程团队进行版本迭代、问题分流与质量评估。
  • Automated the test flow with Python tooling, cutting manual effort and human error.用 Python 工具自动化测试流程，降低人工投入与人为误差。

Open Source开源

GitHub ActivityGitHub 活跃度

A year of commits, reviews, and late-night refactors.一年的提交、评审，和深夜重构。

1,839 contributions in the last year 过去一年共 1,839 次贡献

View profile查看主页
JunJulAugSepOctNovDecJanFebMarAprMayJun
Less少 More多