← Enter the castle

The concise portfolio

Zhihong (Cody) Jiang

ML systems, gameplay AI, and engineering projects. A collection of work in code, research, and small worlds.

About

STUDY / ABOUT

Cody Jiang

Zhihong Jiang / 姜知泓. I work on ML systems, gameplay AI, and software people can use.

Portrait of Cody Jiang from the existing portfolio.
From the existing portfolio

Background

My existing portfolio lists an NYU M.S. in Computer Science (expected December 2026), following an NYU B.S. in Computer Science with minors in Applied Mathematics and Game Design (2021–2024). Current education and employment details are awaiting an update.

Away from the screen

I paint and kitbash miniatures, build small worlds, and share my space with JiJi, my black shorthair cat.

Why a castle?

Each room gives a different part of my work a place of its own. You can explore slowly, or head straight to a project, the CV, or my contact details.

Experience

STUDY / ARCHIVE

Background & Experience

Experience preserved from the existing portfolio. Current roles, dates, attribution, and results should be reviewed against an updated CV.

Gameplay / AI Systems Developer · AbyssalPact

Listed from April 2025 in the previous portfolio. Enemy-turn execution, scoring, target selection, summon placement, fallback handling, extensible AI tags, and runtime debugging in Unity, C#, and XLua. Current status awaits confirmation.

NYU Deep Learning course · September–December 2024

JEPA sequential prediction course project: recurrent image/action modeling, training workflow, representation regularization, and probe-based evaluation.

AI-powered interview simulation · October 2023–March 2024

Engineering work on NLP and audio/video analysis for structured interview feedback, using Python, PyTorch, Transformers, Torchaudio, and SpaCy. Collaborator attribution awaits confirmation.

Software Engineering Intern · June–August 2023

Anhui Yuntai Transportation Development Limited. Mobile operations workflows and a C++ reporting module for bus routes, revenue, trip duration, and passenger flow.

CNN application in cancer diagnosis · June–August 2021

The previous portfolio describes model training, inference integration, UI, and NLP support. Institutional attribution, publication details, and numerical results are awaiting supporting materials and are not asserted here.

Projects

SELECTED WORK / GAME AI

Gameplay AI for AbyssalPact

Enemy AI for a turn-based card game built with a Unity, C#, and XLua hybrid runtime.

Gameplay recording and actual AI/debugging screenshots to be added.

The system

My work covers the enemy AI architecture and its runtime behavior. Extensible AI tags and metadata provide a foundation for different enemy behaviors.

What I built

I built the enemy-turn execution pipeline, including scoring, target selection, summon placement, and fallback handling, with runtime parameters and debugging safeguards.

Materials to be added
  • Demo / release link
  • Approved source-code link
  • Current role, dates, and team attribution

SELECTED WORK / DATA & SOFTWARE

Steam Review Radar

Platform behavior and signal-quality analysis, built around Steam reviews and a large-scale data pipeline.

Actual dashboard screenshots and data visualizations to be added.

The system

A large-scale pipeline brings Steam review data into a workflow for examining platform behavior and the quality of review signals.

Approach

The analysis uses Negative Binomial modeling and XGBoost, connecting statistical methods and machine learning with the data pipeline.

Materials to be added
  • Demo and code links
  • Dataset scope, personal contribution, and verified findings

GALLERY / DATA ENGINEERING

NVIDIA Options Market Pipeline

An options-market pipeline and visualization project listed in the existing portfolio.

Actual pipeline diagram and visualization screenshots to be added.

Tools

Python, Pandas, D3.js, and ETL. Dataset dates, current availability, and project outcomes have not yet been supplied.

Materials to be added
  • Demo and code links
  • Dataset dates, personal contribution, and verified results

GALLERY / DATABASE & BACKEND

Airline Ticket Booking System

A Flask and SQL coursework project for airline ticket reservations, with customer and airline-staff workflows.

Running Flask application screenshots to be supplied. The static template is not presented as a live demo.

Explore the source

The original Flask application, SQL schema, seed data, and test scenarios remain in this repository. GitHub Pages cannot run the Python server. The archived template is available for inspection; it is not a working booking service.

GALLERY / COURSEWORK ARCHIVE

Web & Interactive Studies

Web experiments and response sites from the original portfolio. The original addresses remain available.

The original CSS coffee cup rendered in a browser.
CSS coffee cup · actual coursework page

An archive of small experiments

Front-end coursework includes visual essays, CSS illustration, generative sketches, and interactive storytelling.

Research

RESEARCH / NYU COURSE PROJECT

JEPA World Model

An end-to-end implementation and training workflow for future latent-state prediction, developed for the NYU Deep Learning course, September–December 2024.

Model diagram, prediction video, and evaluation plots to be added.

Approach

The recurrent variant combines image sequences with action inputs to predict future latent states. The work spans model implementation and the training workflow.

Training & evaluation

The workflow uses VICReg/BYOL-style regularization constraints and probe-based evaluation to examine the learned representations. This is a course project.

Materials to be added
  • Course report / poster
  • Approved code repository
  • Experiment settings and verified results

RESEARCH / SOFTWARE ENGINEERING

AI-Generated Code & Technical Debt

A system for analyzing technical debt in AI-generated code, combining code analysis with a practical data-processing workflow.

Actual analysis dashboard or report screenshots to be added.

The system

The AI-Generated Code Technical Debt Analysis System brings static analysis and visualization together to examine technical debt in generated software.

Approach

The project uses Python and Dask for processing, with Git and Docker in the workflow. Static analysis provides material for inspection and visualization.

Materials to be added
  • Demo and code links
  • Study method, personal contribution, and verified findings

Workshop

WORKSHOP / GAME DEVELOPMENT

Games & Playable Experiments

Game projects you can open and explore, alongside my gameplay AI work.

Actual Project 150 browser build, captured locally.
Project 150 · actual browser build
Actual Contra-like arcade browser build, captured locally.
Contra-like arcade · actual browser build
Silent, unedited recording of the original Contra-like browser build: title menu and the opening single-player platform scene.

Try a build

Open Project 150 or a Contra-like arcade game with 1P / 2P modes. The original browser builds remain available at their existing addresses. These older games are best played with a desktop keyboard.

More games

My itch.io page is another entrance to this part of my work. AbyssalPact also has its own exhibit in the Gallery.

Starting these archived builds

Project 150: click the game and hold Space briefly to begin. The Contra-like game: click the game, press 1 or 2 to select a player mode, then hold Space briefly. These GameMaker builds read keyboard input on game frames.

Materials to be added
  • Original source repositories
  • Complete keyboard controls guide

WORKSHOP / PAINTING & KITBASHING

The Miniature Bench

Small worlds made by hand: miniature painting, kitbashing, and the details that make a model feel personal.

Photographs of Cody’s own miniatures: finished pieces, details, and work in progress.

On the bench

Warhammer miniatures are part of my creative life, from painting Ultramarines to modifying models and building terrain.

The collection

This space is intended for finished pieces, close-up details, and work in progress, with room to show how each model takes shape.

About the imagery

The castle and its rooms are concept illustrations, not project screenshots. Scene artwork is illustrative; the finished gallery will use photographs of my own miniatures.

Materials to be added
  • Original model photographs
  • Model names, dates, materials, and process notes

WORKSHOP / STUDIO COMPANION

Meet JiJi

JiJi is my black shorthair cat and studio companion.

An actual photograph of JiJi to be added. The illustrated cat is concept art.

A personal corner

There is room in the castle for the life around the projects, too. This small corner belongs to JiJi.

Back to the workshop

The game projects and miniature bench are just a few steps away.

CV & resumes

STUDY / BACKGROUND & EXPERIENCE

Curriculum Vitae

A direct route to my academic and professional background.

February 2026 edition

This link opens the February 2026 edition of the CV hosted on my existing website.

Areas of focus

My background brings together computer science study at NYU and project work in ML systems, gameplay AI, and software engineering.

Materials to be added
  • A newer CV and current SDE / DS / MLE resumes, if available

Contact

Album

STUDY / PHOTO ALBUM

Off the Resume

A few photographs preserved from the existing portfolio.

GDC photograph
From the existing portfolio · GDC photograph
Graduation photograph
From the existing portfolio · Graduation photograph
Life photograph 1
From the existing portfolio · Life photograph 1
Life photograph 2
From the existing portfolio · Life photograph 2