Nien-Tsu Li — Physical Design / APR & AI

👤

關於我

Professional Profile

職涯摘要： 具軟體開發（Python / Flask / SQL）與 AI 研究（IEEE 期刊）背景；現於 Synopsys 擔任實體設計（Physical Design）與 APR flow 之應用支援，工作重心為 Floorplan 設計與 Physical Cell Placement，並常態參與跨國溝通、R&D／客戶間技術銜接與 Issue Triage。

At a glance: Former software developer & IEEE AI researcher who transitioned into EDA physical design. Now a Synopsys APR application engineer specialising in floorplan design and physical cell placement — with a track record of bridging R&D, AE, and customers across advanced nodes and international teams.

⚡

Physical Design / APR flow engineer

Physical Design / APR Flow Engineer

Synopsys FC / ICC2 ｜現職核心

Synopsys FC / ICC2 ｜ Current Role

於完整 APR flow 中以 Floorplan 規劃與 Physical Cell Placement 為主要工作範疇，在 advanced node 客戶端進行問題分類（Triage）、方法論調整與工具議題追蹤，並與海外 R&D 對齊。

Primary focus within the full APR flow: floorplan design and physical cell placement. Triaging field issues, aligning methodology with advanced-node constraints, and coordinating solutions with overseas R&D teams.

🧬

AI 病理影像研究者

AI Pathology Researcher

IEEE TAI 期刊｜ ISCAS · IJCAI

IEEE TAI Journal ｜ ISCAS · IJCAI

研究聚焦數位病理切片影像：語意分割、Source-Free Domain Adaptation、 Knowledge Distillation 與半監督學習，成果收錄於 IEEE Transactions on Artificial Intelligence。

Research in histopathological image analysis: semantic segmentation, source-free domain adaptation, knowledge distillation, and semi-supervised self-training. Published in IEEE Transactions on Artificial Intelligence.

💻

軟體工程基底

Software Engineering Foundation

Python · C/C++ · Tcl ｜持續精進

Python · C/C++ · Tcl ｜ Active Self-Study

Python 端到端 AI pipeline 與後端（Flask/SQL）；Tcl 腳本自動化 EDA 驗證流程；每日精進 C/C++ 與 Linux Kernel Driver，具備從 high-level script 到韌體層的系統觀。

Python end-to-end AI pipelines and backend (Flask/SQL); Tcl automation for EDA validation. Daily self-study in C/C++ and Linux Kernel/Driver builds complementary system-level understanding alongside high-level scripting.

💼

工作經歷

Work Experience

Synopsys 新思科技

資深應用程式工程師（Sr. Application Engineer）

Senior Application Engineer

2024 年 8 月 — 至今｜台北市

Aug 2024 – Present ｜ Taipei, Taiwan

● 現職 ● Current

EDA 工具實作（FC / ICC2）：主要支援工具為 Fusion Compiler（FC）與 ICC2，協助客戶執行 APR digital implementation，對齊製程規範與設計環境約束。
EDA tooling (FC / ICC2): Primary tools are Fusion Compiler (FC) and ICC2; support customers with APR digital implementation, aligning process design rules and environment constraints.
Floorplan 設計專業：制定 floorplan 規劃方案，整合面積、時序、Power Domain 與繞線可行性需求；在 floorplan-aware 約束層面協助客戶完成設計收斂。
Floorplan design expertise: Define floorplan strategies integrating area, timing, power domains, and routability requirements; guide customers to achieve design closure under floorplan-aware constraints.
Physical Cell Placement 設計：調校 standard / physical cell placement 收斂策略，協助客戶在 PPA（Power / Performance / Area）之間做出有根據的 trade-off 決策。
Physical cell placement: Tune standard/physical cell placement convergence strategies and help customers make informed PPA (Power / Performance / Area) trade-off decisions.
多客戶 × Advanced Node：在職期間同時支援多家客戶，涵蓋多個 advanced technology node 之 methodology 共同開發，依客戶製程規範調整腳本與驗證流程。
Multi-customer · Advanced nodes: Concurrently support multiple customers across multiple advanced technology nodes, co-developing methodology and adapting scripts/flows to each customer's process rules.
Issue Triage 與問題排解：將現場問題精確分類為 usage error、methodology gap 或工具潛在 bug；統整重現環境、推進驗證，提供 workaround 與長期修法建議。
Issue triage & resolution: Accurately classify field issues as usage errors, methodology gaps, or potential tool bugs; consolidate reproduction environments, drive verification, and deliver workarounds plus long-term fix recommendations.
跨國 / 跨部門協作：定期與海外 R&D 團隊會議對齊技術方向；在 R&D ↔ AE/FAE ↔ 客戶 之間擔任技術訊息橋樑，將現場語意轉為可追蹤 issue 規格並完整回饋給用戶端。
Cross-functional / international collaboration: Regular alignment meetings with overseas R&D teams; serve as the technical bridge between R&D ↔ AE/FAE ↔ customers, converting field context into traceable issue specs and communicating conclusions back to end users.
Tcl / Python 自動化：撰寫設計規範 checker、批次驗證與 sign-off 前報表腳本，降低重複性人工除錯成本。
Tcl / Python automation: Develop design-rule checkers, batch validators, and pre-sign-off report scripts to reduce repetitive manual debug effort.

Fusion Compiler (FC) ICC2 Floorplan Design Physical Cell Placement APR Flow Advanced Tech Nodes Issue Triage Overseas Collaboration R&D-AE-Customer Bridge Tcl Python Automation

Synopsys 新思科技

Application Engineer（實習）

Application Engineer Intern

2023.07 – 2023.09 ｜台北｜ 2 個月完成 3 個專題

Jul – Sep 2023 ｜ Taipei ｜ 3 projects completed in 2 months

ICC2 Floorplan Placement & Routing Tcl Python

＋

於 ICC2 環境中以 Tcl + Python 獨立完成三個工具開發專題：Boundary Checker（設計規範符合性驗證）、Random Floorplan Generator（規則化版圖自動生成）、Global Routing / Layer Checker（繞線層合規驗證）。
Independently completed 3 tool-development projects in ICC2 using Tcl + Python: Boundary Checker (design-rule compliance), Random Floorplan Generator (rule-based auto-layout), Global Routing / Layer Checker (routing layer validation).
參與後端 APR 實務：直接接觸 Placement & Routing 流程，奠定現職 EDA APR 的直接基礎。
Hands-on APR practice in Placement & Routing flows — direct foundation for current EDA role.

ICC2 Floorplan Placement & Routing Tcl Python

國立成功大學（NCKU）

National Cheng Kung University (NCKU)

研究助理 — AI 病理影像

Research Assistant — AI Pathology

2023.07 – 2024.07 ｜台南｜ IEEE ISCAS 2024 · IJCAI 2024

Jul 2023 – Jul 2024 ｜ Tainan ｜ IEEE ISCAS 2024 · IJCAI 2024

AI Pathology Domain Generalization Semi-Supervised Learning PyTorch

＋

研究主軸：數位病理切片影像，包含肝細胞核分割（Domain Generalization）、半監督語意分割（Semi-Supervised Self-Training）、跨域一致性對抗背景擾動。
Research in digital pathology: hepatocyte nucleus segmentation (domain generalization), semi-supervised semantic segmentation (self-training), and cross-domain consistency against background perturbation.
完成 IEEE ISCAS 2024 與 IJCAI 2024 論文撰寫，以 PyTorch 設計完整訓練 / 評估 pipeline。
Authored IEEE ISCAS 2024 and IJCAI 2024 papers; designed full training/evaluation pipeline using PyTorch.

AI Pathology Domain Generalization Semi-Supervised Learning Knowledge Distillation PyTorch Python

AUO Corporation（友達光電）

軟體工程師（實習）

Software Engineer Intern

2022.06 – 2022.08 ｜台中｜ 獨立完成全廠 AI 建模平台

Jun – Aug 2022 ｜ Taichung ｜ Built factory-wide AI platform independently

Python Flask CNN / GAN SQL Full-Stack

＋

獨立開發供全廠使用的 AI 建模網頁平台：整合非監督式、少樣本、標準模型與平行運算伺服器，具備完整 end-to-end 系統設計能力。
Independently built a factory-wide AI model-building web platform: unsupervised, few-shot, and standard models with parallel-processing servers — full end-to-end system design.
技術棧：Python、Flask、TensorFlow / PyTorch、CNN、GAN、SQL，完成前後端及資料庫的完整串接。
Stack: Python, Flask, TensorFlow / PyTorch, CNN, GAN, SQL — full-stack integration of backend, model serving, and database.

Python Flask CNN / GAN SQL Full-Stack Development

東海大學 · Ai4kids 講師

Tunghai University · Ai4kids Instructor

研究助理＆ AI 課程講師

Research Assistant & AI Instructor

2020.06 – 2022.07 ｜台中｜ 多所大學 AI 講師 · EPA / IoT 計畫

Jun 2020 – Jul 2022 ｜ Taichung ｜ AI Instructor · EPA / IoT Projects

Technical Teaching Computer Vision IoT Jetson Nano

＋

EPA 計畫：以 AI 影像辨識建立黑白煙不透明度即時量測系統。
EPA project: Real-time AI image-recognition system for black/white smoke opacity measurement.
科技部計畫：整合 IoT + Mobile App 開發高齡照護健康管理平台（Jetson / Raspberry Pi / MQTT）。
MOST project: IoT + mobile app senior care health management platform (Jetson / Raspberry Pi / MQTT).
Ai4kids 講師：受多所高中與大學邀請，教授深度學習與 AI 實作（影像分類 / 偵測），培養科普溝通與技術表達能力。
Ai4kids instructor: Invited by high schools and colleges to teach deep learning and AI implementation — building technical communication skills.

Python Computer Vision IoT Technical Teaching Jetson Nano

🛠

技術能力

Technical Skills

⚡ EDA / APR

現職核心

Core Expertise

Fusion Compiler (FC) ICC2 Floorplan Design Physical Cell Placement APR Flow Advanced Technology Nodes PPA Optimization Issue Triage Methodology Support Customer Application Support Tcl Scripting Cross-team Collaboration

🐍 Software
Python · AI

學術 & 工程背景

Research & Engineering

Python AI Pathology Image Recognition Semantic Segmentation PyTorch TensorFlow Domain Adaptation Knowledge Distillation Semi-Supervised Learning Flask SQL CNN / GAN OpenCV · CUDA

🔧 C / C++
韌體 · 系統

🔧 C / C++
Firmware · Systems

持續進修（C/C++ · 系統）

Daily Self-Study

C Language C++ Pointer / Memory Management Bitwise Operations Linux-style Linked List malloc / calloc Linux Kernel Driver / Firmware Jetson Nano · Raspberry Pi IoT · MQTT

📄

學術發表

Publications

IEEE
TAI

Domain-Centroid-Guided Progressive Teacher-based Knowledge Distillation for Source-Free Domain Adaptation of Histopathological Images

IEEE Transactions on Artificial Intelligence (TAI) · 2023 · 期刊論文 / Journal

數位病理切片跨域適應：以教師模型 Progressive 蒸餾為核心，在無源域資料的情境下達成跨院所影像遷移。

Addresses domain shift in histopathology without source-domain data access; uses progressive teacher-based knowledge distillation guided by domain centroids.

→ View on IEEE Xplore ↗

IEEE
ISCAS
2024

Domain Generalization with Anti-background Perturbation Consistency and Texture Reduction Ensemble Models for Hepatocyte Nucleus Segmentation

IEEE ISCAS 2024 · 國際會議 / Conference

肝細胞核分割之跨域泛化：對抗背景擾動一致性損失 + 紋理縮減 Ensemble，提升在未見染色域的穩健性。

Hepatocyte nucleus segmentation with improved cross-domain robustness via anti-background perturbation consistency and texture-reduction ensemble strategy.

IJCAI
2024

Reliable Self-Training with Class Group Feature Alignment for Semi-Supervised Semantic Segmentation

IJCAI 2024 · 國際會議 / Conference

以類別群組特徵對齊提升半監督語意分割中自訓練的可靠性，降低偽標籤雜訊對模型收斂的影響。

Improves pseudo-label reliability in semi-supervised semantic segmentation through class group feature alignment, reducing noise from incorrect pseudo labels during self-training.

🎓

學歷

Education

M.S. 國立成功大學（NCKU） NCKU 電機工程 · AI 病理影像 EE · AI Pathology 2022–2024

·

B.S. 東海大學 Tunghai University 電機工程 EE 2018–2022

🎬

內容創作與部落格

Content Creation & Technical Writing

▶

YouTube 精選教學影片

YouTube Featured Tutorial

@hardenli6973 · 1,000+ Views

🔥 精選 / Featured 1,000+ Views OpenCV · CUDA

【OpenCV CUDA】How to Build OpenCV GPU with CUDA on Windows

↑ 點擊縮圖至 YouTube 觀看

↑ Click thumbnail to watch on YouTube

【OpenCV CUDA】Windows GPU 加速完整教學

How to Build OpenCV with GPU/CUDA on Windows

從 CMake 設定到 Visual Studio 編譯 的完整手把手教學。影片累積 逾 1,000 次觀看，可作為在 Windows 上建置 OpenCV GPU 環境的參考流程。

Step-by-step from CMake to Visual Studio compilation. 1,000+ views — a practical reference for building OpenCV with GPU support on Windows.

▶ Watch on YouTube ↗
→ View All Videos ↗

📝

C / C++ 技術部落格

C / C++ Technical Blog

HackMD · Daily Self-Study

追蹤 Jserv（黃敬群教授）Linux Kernel Driver 課程，每天 4 小時精進。公開筆記：指標 / 記憶體、Linux Linked List、bitwise、malloc/calloc、韌體程式設計、C++ 進階主題。

Following Jserv's (Prof. Jim Huang) Linux Kernel Driver course, 4 hrs/day. Public notes: pointer/memory, Linux linked list, bitwise ops, malloc/calloc, firmware, advanced C++.

→ C Language Notes ↗ → C++ Blog ↗

⚙