available for work

Sunny Chen

Software Developer

I like building.

// F1 GARAGE11 TEAMS
THE PERSONAL PORTFOLIO OF A SOFTWARE DEVELOPERMONTRÉAL, QUÉBEC · INDEPENDENT PORTFOLIO

Sunny Chen& Co.

Independent thinking.
Compounding curiosity.
SOFTWARE, SYSTEMS & THE OCCASIONAL MARKET OBSESSIONPERSONAL EDITION / NO. 01

A NOTE FROM THE DEVELOPER

Good things
take building.

I’m Sunny, a developer in Montréal. I follow my curiosity into software, markets, and the systems that connect them.

This is a collection of what I’ve built, what I’m learning, and where I’m headed next.

Read the project index
FIELD NOTES001 / MARKET DATA
ON TURNING CURIOSITY INTO CODE

A closer look
at the markets.

A live engine that recomputes options-implied support and resistance every five minutes. Built with Python, PyTorch and Azure.

Inside the project
SCROLL TO THE FUNDAMENTALS CODE / CURIOSITY / CRAFT
01 / A LITTLE CONTEXT

About Me

I'm a Software Engineering Co-op student at McGill and a developer in Montreal with 5+ years of building, from web apps in React, Next.js, and TypeScript to backend services in Python, Java, and C#/.NET. I care about shipping things that hold up under real use, not just things that work.

Most of what I build comes out of whatever I'm into — finance led to the Options Market Analytics Engine (a live options-analytics system on Azure), F1 has me deep in an ML race-prediction model, and my internship at Galerie L’Original had me shipping a RAG image verifier, a View in Room AR feature, and a SAM-based auto-crop pipeline. Half my projects start as “I wonder if I could build that.”

McGill University
B.Eng. Software Engineering Co-op · Aug 2025 – May 2029
Vanier College
DEC Computer Science & Mathematics · Aug 2023 – Jun 2025

0

Projects Shipped

0

Game Jam 2024

0

Languages

02 / TRACK RECORD

Work Experience

POSITION / 01Jun 2026 – Oct 2026Montreal, QC

Full Stack Developer Intern @ Galerie L’Original

Shipped features on two production Next.js/TypeScript sites (React, MySQL) in a 20+ developer agile team, contributing through code reviews and sprint planning. Replaced manual image quality review with a RAG verifier (Gemini, pgvector), shipped only once it matched human reviewer decisions. Scoped and built View in Room AR from spec to release, generating true-to-scale 3D models (three.js, GLB/USDZ) for 421 artworks in 16 minutes. Developed a SAM 2.1 auto-crop pipeline for 3,600+ listings; fine-tuned it in PyTorch, raising boundary IoU from 0.943 to 0.968.

TypeScriptNext.jsReactMySQLPostgreSQLPrismaGeminiRAGthree.jsPyTorchSAM 2.1
POSITION / 02Jun 2025 – Oct 2025Montreal, QC

Software QA Analyst @ GBT Solutions

Identified 15+ edge-case defects before production by testing restaurant POS software against Revenu Québec fiscal standards. Accelerated fixes by debugging failures down to precise reproduction steps and root-cause analysis for each defect. Created compliance documentation with step-by-step workflows and visual evidence for external auditors.

QAPOSRoot-Cause AnalysisComplianceRevenu Québec
career — -zsh — 80×24

➜ career git log --work-experience

commit 7f153b5 (HEAD -> main)

Author: Sunny Chen

Date: Jun 2026 – Oct 2026

 

Full Stack Developer Intern @ Galerie L’Original · Montreal, QC

 

Shipped features on two production Next.js/TypeScript sites (React, MySQL) in a 20+ developer agile team, contributing through code reviews and sprint planning.

Replaced manual image quality review with a RAG verifier (Gemini, pgvector), shipped only once it matched human reviewer decisions.

Scoped and built View in Room AR from spec to release, generating true-to-scale 3D models (three.js, GLB/USDZ) for 421 artworks in 16 minutes.

Developed a SAM 2.1 auto-crop pipeline for 3,600+ listings; fine-tuned it in PyTorch, raising boundary IoU from 0.943 to 0.968.

 

tags: TypeScript Next.js React MySQL PostgreSQL Prisma Gemini RAG three.js PyTorch SAM 2.1

link: https://loriginal.org

link: https://artur.art

 

commit 799ab49

Author: Sunny Chen

Date: Jun 2025 – Oct 2025

 

Software QA Analyst @ GBT Solutions · Montreal, QC

 

Identified 15+ edge-case defects before production by testing restaurant POS software against Revenu Québec fiscal standards.

Accelerated fixes by debugging failures down to precise reproduction steps and root-cause analysis for each defect.

Created compliance documentation with step-by-step workflows and visual evidence for external auditors.

 

tags: QA POS Root-Cause Analysis Compliance Revenu Québec

➜ career  

Choose a stop to explore each role.

Career circuit / Nürburgring 24H

Galerie L’Original

Full Stack Developer Intern

Jun 2026 – Oct 2026 · Montreal, QC

Session highlights

  • Shipped features on two production Next.js/TypeScript sites (React, MySQL) in a 20+ developer agile team, contributing through code reviews and sprint planning.
  • Replaced manual image quality review with a RAG verifier (Gemini, pgvector), shipped only once it matched human reviewer decisions.
  • Scoped and built View in Room AR from spec to release, generating true-to-scale 3D models (three.js, GLB/USDZ) for 421 artworks in 16 minutes.
  • Developed a SAM 2.1 auto-crop pipeline for 3,600+ listings; fine-tuned it in PyTorch, raising boundary IoU from 0.943 to 0.968.

Technical setup

TypeScriptNext.jsReactMySQLPostgreSQLPrismaGeminiRAGthree.jsPyTorchSAM 2.1
loriginal.orgartur.art
03 / SELECTED WORK

The project index.

A running record of experiments, useful tools, and things made out of curiosity.

SC / 01Shipped

CrowdCheck

Built a full-stack, real-time crowdedness tracker so McGill students can check gym zones and library floors before going. Designed a RESTful .NET API with rate-limited voting and time-decayed aggregation for accurate crowd estimates. Used Dapper and DbUp for raw SQL data access and versioned PostgreSQL migrations; deployed through a CI/CD pipeline.
C#.NETAngularTypeScriptPostgreSQLDapper
SourceExplore project
SC / 02Shipped

Ideal Gas Law Simulator

Built an Ideal Gas Law simulator in Java/JavaFX to model the relationship between variables in real time. Implemented interactive visualizations and user controls to dynamically adjust variables and observe simulation outcomes. Led a team of 3 to accomplish the project with tight deadlines.
JavaJavaFXScene BuilderGit
SourceExplore project
SC / 03Shipped

Vanier Robotics Team Website

A Plants vs. Zombies themed website built for the Vanier College Robotics Team. Designed and developed an engaging, game-inspired interface to showcase the team and attract new members. Built with Vue.js for a dynamic and interactive user experience.
Vue.jsJavaScriptCSSHTML
Explore project
SC / 04Shipped

Université de Montreal's Game Jam 2024

Collaborated as part of a dynamic team of five to achieve first place in a competitive Game Jam competition. Developed and implemented innovative game concepts within tight time constraints, resulting in recognition for excellence. Led the conceptualization and design of a game project.
C#UnityGit
Explore project
SC / 05Shipped

PhysiPlay – 2D Physics Simulator

Developed a 2D physics simulation engine in Java and JavaFX, modeling pendulums, springs, collisions, and momentum. Applied object-oriented programming (OOP) and design patterns to build modular and scalable simulation modules. Collaborated in a team of 4 using Agile methodologies (Scrum, sprint planning, stand-ups) to ensure efficient project delivery. Utilized Git/GitHub for version control, implementing branching strategies and code reviews to maintain code quality.
JavaJavaFXScene BuilderGit
SourceExplore project
SC / 06Shipped

Gmail Internship Scanner

LLM-powered tool that scans a Gmail inbox to surface internship-related emails and extract structured details like company, role, and status. Integrates with the Gmail API to fetch messages and runs a local Qwen model via Ollama for private, offline classification and extraction. Designed to keep all email content on-device while automating the tedious work of tracking internship applications.
PythonGmail APIOllamaQwenLLM
SourceExplore project
SC / 07In development

Options Market Analytics Engine

Built a live analytics engine on Azure that recomputes options-implied support and resistance levels every 5 minutes. Backtested a hedging strategy on 4.7 years of 5-min market data, cutting P&L variance 92% and tail losses 76%. Validated that the engine’s signal improves volatility forecasts (p = 0.005) using walk-forward and permutation tests.
PythonPyTorchSpring BootMicrosoft AzurePostgreSQL
SourceExplore project
SC / 08In development

F1 Race Prediction Model

Machine learning model to predict Formula 1 race outcomes based on historical data, driver performance, and track conditions. Utilizing Python, Pandas, Scikit-learn, and TensorFlow to preprocess data, engineer features, and train predictive algorithms. Aiming to provide insights for teams and enthusiasts to enhance race strategies and fan engagement.
PythonPandasScikit-learnTensorFlow
SourceExplore project
SC / 09In development

Git Scribe – Commit Documentation Generator

CLI tool that reads a repository's Git history and generates human-readable documentation and changelogs from commit messages and diffs. Uses a local Qwen model via Ollama to summarize changes, group related commits, and produce release-ready notes without sending code off-machine. Built in Python to streamline documentation for solo projects and small teams.
PythonGit CLIOllamaQwenLLM
SourceExplore project

Showing 9 of 9 projects

My Projects

CrowdCheck

Built a full-stack, real-time crowdedness tracker so McGill students can check gym zones and library floors before going. Designed a RESTful .NET API with rate-limited voting and time-decayed aggregation for accurate crowd estimates. Used Dapper and DbUp for raw SQL data access and versioned PostgreSQL migrations; deployed through a CI/CD pipeline.

C#.NETAngularTypeScriptPostgreSQLDapper
Ideal Gas Law Simulator

Built an Ideal Gas Law simulator in Java/JavaFX to model the relationship between variables in real time. Implemented interactive visualizations and user controls to dynamically adjust variables and observe simulation outcomes. Led a team of 3 to accomplish the project with tight deadlines.

JavaJavaFXScene BuilderGit
Vanier Robotics Team Website

A Plants vs. Zombies themed website built for the Vanier College Robotics Team. Designed and developed an engaging, game-inspired interface to showcase the team and attract new members. Built with Vue.js for a dynamic and interactive user experience.

Vue.jsJavaScriptCSSHTML
Université de Montreal's Game Jam 2024

Collaborated as part of a dynamic team of five to achieve first place in a competitive Game Jam competition. Developed and implemented innovative game concepts within tight time constraints, resulting in recognition for excellence. Led the conceptualization and design of a game project.

C#UnityGit
PhysiPlay – 2D Physics Simulator

Developed a 2D physics simulation engine in Java and JavaFX, modeling pendulums, springs, collisions, and momentum. Applied object-oriented programming (OOP) and design patterns to build modular and scalable simulation modules. Collaborated in a team of 4 using Agile methodologies (Scrum, sprint planning, stand-ups) to ensure efficient project delivery. Utilized Git/GitHub for version control, implementing branching strategies and code reviews to maintain code quality.

JavaJavaFXScene BuilderGit
Gmail Internship Scanner

LLM-powered tool that scans a Gmail inbox to surface internship-related emails and extract structured details like company, role, and status. Integrates with the Gmail API to fetch messages and runs a local Qwen model via Ollama for private, offline classification and extraction. Designed to keep all email content on-device while automating the tedious work of tracking internship applications.

PythonGmail APIOllamaQwenLLM
04 / TOOLS OF THE TRADE

Skills & Technologies

Languages
Java
Python
TypeScript
JavaScript
C#
C
SQL
Bash
Frontend
React
Next.js
Angular
Tailwind CSS
HTML5
CSS3
DevOps & Tools
Git
GitHub
GitLab
Docker
Azure
Linux
JUnit
Maven
Gradle
Backend
Spring Boot
.NET
Node.js
Databases
PostgreSQL
MySQL
Machine Learning
PyTorch
scikit-learn
05 / CORRESPONDENCE
Have something
in mind?

Let's Connect

I'm always open to discussing new projects, opportunities, or just having a chat about technology.

EmailGitHubLinkedIn