STEVEN KANG

Computer Science · University of Michigan

// PLAYER PROFILE

About Me

Computer Science student at the University of Michigan working in data science and NLP research. Enjoy building software tools and looking to learn more about systems programming.

Current roles have provided experienced in building web scraping pipelines, LLM evaluation frameworks, and open-source CLI tools. Have conducted research across multiple departments at UMich, from institutional research data pipelines to NLP-powered policy analysis.

3 Research Roles

2 Projects Shipped

steven@umich:~$ whoami

Steven Kang

steven@umich:~$ cat major.txt

B.S. Computer Science

steven@umich:~$ _

// SELECT LEVEL

Experience

LVL 3

Data Science Research Assistant

Rackham Graduate School · U-M Institutional Research

JUN 2026 — PRESENT ★ CURRENT

Built web scraping pipelines using Python and BeautifulSoup to extract and link PhD student and advisor records across institutional and public academic sources, creating a unified dataset for mentorship research
Integrated the Google Gemini API (with URL-context tooling) to classify and extract structured data (name, title, lab, advisors, department) from 1000+ unstructured student profile pages

Python BeautifulSoup Web Scraping Data Pipelines

LVL 2

Research Assistant

School of Information · U-M — Joshua Ashkinaze (supervisor)

MAY 2026 — PRESENT ★ CURRENT

Built a quiz-based evaluation framework to test whether stakeholder descriptions from an LLM extraction-entailment model captured key ballot information, achieving 95\% answer accuracy across 50 measures
Refactored an LLM pipeline that extracts structured information (key entities, relationships, and supporting evidence) from text to have configurable options for how much context the system uses as input
Ran statistical comparisons of output quality across different configurations using automated metrics

Python LiteLLM NLP LLM Evaluation Statistical Analysis

LVL 1

Research Assistant

Gerald R. Ford School of Public Policy · U-M — Dr. Devin Judge-Lord (supervisor)

SEP 2025 — APR 2026 ✓ COMPLETE

Constructed a crosswalk of 800,000+ unique corporations using regular expressions and token-based string-matching algorithms in Python to identify corporate aliases in 900,000+ records across multiple datasets
Built an open-source software package in R for users to search corporation aliases in semi-structured text
Deployed a full-stack web application using Next.js and PostgreSQL to accept alias suggestions from users

Python R Next.js PostgreSQL Regex

📂 View on GitHub →

// INVENTORY

Skills & Stats

⚔️ Languages

Python LVL 9

C++ LVL 7

Java LVL 6

R LVL 7

SQL LVL 7

🛡️ Libraries & Frameworks

📊pandas

🤖LiteLLM

🧠scikit-learn

📈Seaborn

🕸️BeautifulSoup

🗄️SQLAlchemy

🧪GoogleTest

🎒 Tools & Platforms

🔀Git

📓Jupyter

⚡Supabase

▲Vercel

📮Postman

🔧CMake

// QUEST LOG

Projects

🖥️

Warden — macOS Process Manager

Open-source macOS CLI tool using low-level system APIs (libproc) to scan running processes, group by parent application, and rank search results with a custom heap-based algorithm
Configured build system using CMake, validated core logic with unit tests via GoogleTest, and setup installation via Homebrew tap

C++ CMake libproc GoogleTest Homebrew

📂 View on GitHub

📋

Timesheet Management System

Designed a SQLite database schema for a software firm to efficiently store and query time-sheet records
Implemented database models using SQLAlchemy ORM with normalized table relationships
Wrote a data cleaning pipeline to standardize raw time-sheet inputs for automated and consistent formatting

Python SQLite SQLAlchemy Data Cleaning

🏢 Innovation for Impact

// ACHIEVEMENTS UNLOCKED

Education & Leadership

🎓

University of Michigan

Bachelor of Science in Computer Science

Ann Arbor, MI · Expected Graduation: Dec 2028

✦ In Progress ✦

⚡

Business Director

Innovation for Impact — Technology Consulting Group

Apr 2026 — Present

Manage club finances including budgeting, dues collection, reimbursements, and financial reporting
Coordinate administrative tasks such as attendance tracking, roster updates, and meeting room reservations

✦ ACTIVE ROLE ✦

// SIDE QUESTS

Interests

⚽

Soccer

Watching and playing — I would never say no to a pickup game or catching a match on TV, soccer is part of my life.

⚡ Playing 📺 Watching

🏋️

Gym

Consistent gym sessions let me stay active and take a break from coding sessions.

💪 Lifting 🔥 Gains

🌯

Chipotle

The order is non-negotiable:

▸ Bowl, wrap on the side

▸ White rice

▸ Pinto beans

▸ Chicken

▸ Red salsa

▸ Corn

▸ Sour cream

▸ Extra cheese

▸ A little lettuce

▸ No drink

// GAME OVER? NOT YET

Get In Touch

Feel free to contact me through one of the options below

✉️ stevenhk@umich.edu 💻 GitHub 🔗 LinkedIn