About Me

I'm Sheng-Wei (Roger) Fan, a third-year CS student at NTHU who enjoys building software and solving technical problems. Self-driven and quick to learn —— whether it's my own dev env, architecting full-stack systems, or ML research. I turn ideas into working implementations across hackathons, research, and personal projects.

Dev Environment

Dev Environment
Click to enlarge
View Dotfiles Repo

A highly customized terminal-based development environment built with modern tools and extensive automation. Focused on efficiency and aesthetics with Lua-based configurations.

  • •Neovim (Lua): LSP via mason.nvim, telescope.nvim fuzzy finder, GitHub Copilot, neo-tree file explorer, vimtex (LaTeX), leetcode.nvim
  • •Tmux: Session auto-save/restore, custom keybindings (Ctrl+Space prefix), TPM plugin manager
  • •Shell: Zsh + Powerlevel10k prompt + fast-syntax-highlighting + zsh-autosuggestions
  • •Wezterm: GPU-accelerated Rust terminal with Lua configuration and Catppuccin theme
  • •Git Workflow: lazygit.nvim TUI, gitsigns.nvim inline diff, extensive custom keybindings
  • •Automation: One-command setup script with dotfile symlink management (initialization.sh)

Education

National Tsing Hua University

Bachelor of Science in Computer Science

Aug 2023 - Present
  • •GPA: 4.21/4.3 (Top 3% in Department of Computer Science) cert

Key Courses

Algorithms
Data Structures(A+)Competitive Programming(A+)
Systems
Computer Architecture(A+)Operating System(A+)Computer Network(A+)
AI/ML
Machine Learning(A+)Computer Vision(Lab)Reinforcement Learning(Lab)
Software
Software Studio(A+)Computer System Administration(A+)Cyber Security(A+)
Hardware
Logic Design(A+)Hardware Design & Lab(A+)

Experience

Summer Research Intern

Jul 2025 - Aug 2025

Institute of Information Science, Academia Sinica

Taipei, Taiwan
  • •Built cross-attention heatmaps (layer-wise/head-wise) to identify failure patterns, enabling research team to pinpoint and refine model flaws.
  • •Designed experiments with custom crossing datasets using SAM2, systematically exposing VideoGrain limitations.
PyTorchComputer VisionModel Interpretability

Teaching Assistant

Feb 2025 - Present

Linear Algebra & Introduction to Programming (II), NTHU

Hsinchu, Taiwan
  • •Developed C++ Allegro game template enabling students to apply OOP principles in game development.
  • •Designed and graded assignments, projects, and exams.
C++TeachingCourse Design

Awards

3rd Place, 2026 TSMC IT CareerHack — AAID Challenge
Click to enlarge

3rd Place, 2026 TSMC IT CareerHack — AAID Challenge

TSMC IT CareerHack

Addressed the challenge of single-image outdoor light estimation by conducting comprehensive error analysis and 25+ ablation experiments. Identified and mitigated the model's reliance on global geometric anchors (horizon lines) by introducing Spatial Attention (CBAM) and optimizing backbone architectures. The final EfficientNet-ResNet hybrid model achieved 0.9867 AUC on the SynthStreet dataset.

Feb 2026
1st Place, NSF HDR Scientific-MOOD Hackathon — Imageomics Ecology ML Challenge (Taiwan Local Competition)
Click to enlarge

1st Place, NSF HDR Scientific-MOOD Hackathon — Imageomics Ecology ML Challenge (Taiwan Local Competition)

U.S. National Science Foundation (NSF) Harnessing the Data Revolution (HDR)

Developed ML models to predict drought severity (SPEI index) from NEON ground-beetle specimen images. The challenge emphasizes out-of-distribution generalization, requiring models to perform reliably across unseen ecological sites with different climates and species distributions.

Jan 2026

Academic Achievements Award

National Tsing Hua University

Spring 2024

Skills

Languages

C/C++PythonLuaDartJavaScriptTypeScriptSQLBashLaTeXVerilog

Frameworks

FlutterNext.jsReactTailwind CSSAuth.jsPyTorchScikit-learn

Tools & Technologies

GitDockerFirebaseAWSPostgreSQLPrismaMongoDBNginxPostfixWireGuardBind9NeovimTmuxZshWeztermVivadoAllegro