Data science · NLP · Applied AI

I build data and AI systems that turn complex information into useful decisions.

I’m Stella (Juanru) Zhang, a Cornell Information Science graduate and incoming NYU Data Science master’s student. My work spans human–LLM value alignment, natural-language interfaces for structured data, machine learning, analytics, and software engineering.

View selected work View résumé GitHub ↗ LinkedIn ↗
Cornell University B.A. Information Science, 2026
New York University Incoming M.S. Data Science, Fall 2026
Focus NLP, AI systems, data analysis

Selected work

Projects

Data science · NLP 01

E-Commerce Review Text Mining

Analyzed women’s clothing reviews using a reproducible R and Quarto workflow. Cleaned and transformed text, built TF-IDF features, reduced dimensionality with SVD, explored customer themes with k-means clustering, and evaluated logistic-regression models.

What this demonstrates: end-to-end text analysis, statistical modeling, model evaluation, and clear technical reporting.

  • R
  • Quarto
  • tidytext
  • TF-IDF
  • SVD
Read the full report ↗ View the code on GitHub ↗
Algorithms · Java 02

Sewer Maze Pathfinder

Implemented the decision logic for a maze-navigation game: locate a hidden target, compute an escape route, and collect additional value without exceeding the remaining movement budget.

What this demonstrates: graph traversal, shortest-path reasoning, priority queues, and decision-making under constraints.

  • Java
  • Graphs
  • Shortest paths
  • Priority queues
Review the implementation on GitHub ↗
Systems · OCaml 03

Terminal Chat Server

Built a terminal-based client–server chat application with TCP communication, asynchronous message handling, and broadcast messaging across connected clients.

What this demonstrates: networking fundamentals, concurrency, functional programming, and protocol-oriented system design.

  • OCaml
  • Lwt
  • TCP
  • Networking
View the server and client code ↗

Additional projects

Academic Genealogy Tree Java · trees · recursion · CSV parsing · JUnit Red-Black Tree Set OCaml · balanced trees · OUnit · performance testing Click-a-Dot Game Java Swing · event handling · timers · GUI state

Experience & research

Experience

May 2025–Present

Ithaca, NY

Research Assistant

Cornell Human–LLM Value Alignment Research

  • Designed an experimental framework to study how LLM-generated constructive comments affect the trajectory and rhetoric of online discussions.
  • Conducted large-scale R and SQL analysis of value alignment, cultural homogenization, and model-generated interventions.
  • Co-authored the ACM CHI 2026 paper “LLMs Homogenize Values in Constructive Arguments on Value-Laden Topics” , examining how LLM rewrites can suppress some values while amplifying others in contentious online discussions.

Jun–Oct 2024

Remote, US

Software Engineer

Convoloo — NLP Internship

  • Built a LangChain and Gemini question-answering system that translated natural language into SQL for structured patient-health datasets.
  • Achieved 95% accuracy on basic information-retrieval queries and integrated MySQL and SQLite with AI inference pipelines.
  • Developed conversational retrieval chains and output templates that made generated queries and results easier to inspect and debug.

Jun–Aug 2024

Beijing, China

Data Analyst

Xiaomi — Data Analysis Internship

  • Validated weekly product data for more than 5,000 SKUs across labels, costs, revenue, and inventory status.
  • Built four Power BI dashboards tracking product lifecycles, regional sales, and inventory turnover.
  • Used SQL and Excel to improve reporting workflows across six regions, reducing report-construction time by 30%.
Earlier experience

Software Engineer / Data Analyst

Hisense Transport Statistics Management

Developed debugging protocols, integrated Alibaba Sentinel for request regulation, and supported a cross-functional peer-review process.

Programmer / Data Analyst

Baidu Data Analysis Mentorship

Evaluated XGBoost, Random Forest, Logistic Regression, and Naïve Bayes models on more than 10,000 website visits, achieving over 95% classification accuracy.

Technical toolkit

Skills

Languages

Python, SQL, R, Java, JavaScript, C++, OCaml, HTML, CSS

Data & machine learning

cikit-learn, XGBoost, random forest, gradient boosting, logistic regression, k-means, PCA, cross-validation, hyperparameter tuning, feature engineering, MySQL, SQLite, LangChain, retrieval-augmented generation (RAG), text-to-SQL, prompt engineering, large language models (LLMs)

Analytics and Visualization

pandas, NumPy, Power BI, Tableau, Matplotlib, seaborn, ggplot2, Microsoft Excel, A/B testing, statistical analysis

Developer Tools

GitHub, Jupyter, VS Code, Node.js

Contact

Let’s talk about data, NLP, or applied AI.

I’m interested in research and engineering work where careful analysis and practical software meet.

stllazhn@gmail.com GitHub ↗ LinkedIn ↗