Hi, I'm Leo Yin

AI Engineer & Computer Engineering Student

Building innovative AI solutions and engineering robust systems. Currently focused on AI research at CASIA and pursuing Computer Engineering at the University of Toronto.

Leo Yin - Profile
AI/ML
React
Python
TensorFlow
Scroll to explore

About Me

Computer Engineering Student

I am a Computer Engineering student at the University of Toronto, focusing on the intersection of hardware and software systems, with a particular interest in AI/ML applications.

“Building technologies that make a positive impact on the world.”

Education

University of Toronto

Bachelor of Applied Science and Engineering in Computer Engineering + PEY Co-op

Sep 2023 – Apr 2027

Problem SolvingLeadershipResearchInnovation

AI Researcher

Currently conducting research on Multimodal Large Language Models at CASIA, focusing on developing innovative approaches to improve cross-modal understanding and performance.

Entrepreneur

Founded VolunTrack Org., a non-profit organization with a ML-powered volunteer matching system, serving 1000+ monthly active users and partnering with 100+ global non-profits.

Developer

Skilled in full-stack development with expertise in AI integration, from building ML-powered applications to creating user-friendly interfaces with React, Next.js, and more.

Work Experience

My professional journey through research, entrepreneurship, and development

Feb 2024 - Sep 2024

Research Assistant

The Institute of Automation, Chinese Academy of Sciences (CASIA)Beijing, China
  • Streamlined research on Multimodal Large Language Models (MLLMs) through systematic literature review and novel experimental design, developing methodology for MICCAI 2025 submission
  • Engineered an end-to-end data pipeline for benchmark creation using Python, incorporating automated quality checks and custom annotation tools, resulting in a 75% reduction in processing time and 40% improvement in annotation accuracy
  • Developed a multimodal evaluation framework using PyTorch, benchmarking state-of-the-art MLLMs across multiple dimensions with metrics for accuracy, robustness, and cross-modal consistency
  • Spearheaded weekly technical discussions and authored comprehensive research documentation, facilitating knowledge transfer across interdisciplinary teams of vertical category experts and ML researchers
Jun 2022 - Sep 2024

President/Founder

VolunTrack Org.Toronto, ON
  • Founded and scaled a non-profit organization through a structured four-tier management system, growing the company to 50+ members and establishing partnerships with 100+ global non-profits
  • Architected an ML-powered volunteer matching system using TensorFlow and scikit-learn, achieving 85% matching accuracy and reducing manual matching time by 70%
  • Developed a full-stack volunteer platform with React Native and React.js, integrating MLOps practices and employing a CI/CD pipeline to ensure consistent and robust application builds, serving 1000+ monthly active users
  • Implemented real-time analytics using Google Cloud Platform, Cloud Firestore, and TensorFlow.js, deploying ML models for volunteer engagement prediction and churn analysis across 100+ organizations
  • Established an AI innovation program, leading to 30+ successful ML projects and 10+ conference speaking engagements
Jun 2023 - May 2024

AI Engineering Lead / Co-Founder

YiXing - AI-Driven Personalized Travel PlannerToronto, ON
  • Architected and deployed a production-grade LLM system with fine-tuned GPT-based models, training on 1,000+ curated travel itineraries to achieve an 85% user satisfaction rate based on post-trip feedback
  • Engineered a RAG pipeline processing 5000+ travel documents using LangChain, Pinecone vector DB, and custom chunking strategies, improving travel suggestion accuracy by 35% and reducing waiting time by 90%
  • Developed serverless APIs using AWS Lambda to handle LLM integration, user data management, and travel preference processing for the React Native frontend
  • Led a team of 6, implementing A/B testing and performance monitoring to continuously improve the fine-tuned model
Jan 2025

AI/ML Project Lead

ReassurED - AI-Powered Emergency Care Guidance/NavigatorMontreal, QC
  • Architected an AI-powered healthcare system using Deepseek-v3, React Native (Expo) and integrated real-time hospital data via Firestore, achieving 90% alignment with the standard Emergency Severity Index during a 24-hour hackathon
  • Implemented few-shot learning through medical examples, JSON schema validation, and chain-of-thought reasoning prompt engineering, reducing incorrect classifications by 25% compared to a zero-shot baseline
  • Developed a hybrid recommendation system combining LLM outputs with a weighting algorithm for hospital suggestions, integrating real-time wait times, distance metrics, and facility capabilities, reducing user decision time by 70%
  • Built a scalable FastAPI backend processing 100+ concurrent requests per second, enabling real-time integration between the LLM service and the hospital data system with 99% uptime
Jun 2024 - Present

Web Team Associate

IEEE UofTToronto, ON
  • Developed and maintained full-stack web applications with a React.js frontend and Django backend, enhancing IEEE UofT's main website and event platforms with a responsive, user-friendly design
  • Improved performance by optimizing APIs and caching, reducing load times and enhancing user experience
  • Collaborated within an Agile team to build and deploy new features, ensuring a seamless experience

Projects

Innovative solutions I've built to solve real-world problems

YiXing - AI Travel Planner logo

YiXing - AI Travel Planner

A personalized travel planning platform powered by fine-tuned LLMs, providing custom itineraries based on user preferences.

Next.jsLangChainGPT-4AWS LambdaVector DB

Key Achievements:

  • 85% user satisfaction rate post-trip
  • RAG pipeline with 5000+ travel documents
  • 35% improvement in suggestion accuracy
VolunTrack Platform logo

VolunTrack Platform

A volunteer management platform with ML-powered matching, connecting organizations with volunteers based on interests and skills.

React NativeTensorFlowFirebaseCI/CDMLOps

Key Achievements:

  • 1000+ monthly active users
  • Partnerships with 100+ global non-profits
  • 85% matching accuracy rate
ReassurED

ReassurED Emergency Care Navigator

An AI-powered healthcare guidance system helping users assess symptoms and find appropriate emergency care facilities.

React Native ExpoDeepseek-v3FirestoreFastAPIFew-shot learning

Key Achievements:

  • 90% alignment with Emergency Severity Index
  • 25% reduction in incorrect classifications
  • 70% reduction in user decision time

Skills & Expertise

Technologies and tools I work with

Programming Languages

Python95%
JavaScript90%
C/C++80%
C#75%
HTML/CSS85%
Swift70%
MATLAB75%

Frameworks & Libraries

TensorFlow90%
LangChain85%
OpenCV80%
PyTorch85%
React90%
Node.js85%
Django75%

Cloud & DevOps

AWS85%
Google Cloud80%
Firebase90%
Git95%
CI/CD80%

Development Tools

Vertex AI85%
GitHub95%
Xcode75%
Unity70%
Fusion 36065%
Blender60%
AutoCAD65%

Other Technologies & Tools

Next.jsTypeScriptExpressMongoDBGraphQLReduxFlaskFastAPIDockerKubernetesJenkinsREST APIScikit-learnPandasNumPyNLTKHugging FaceLangchainJIRAFigmaAdobe XDAgileScrumUI/UX Design

Contact Me

Have a question or want to work together? Feel free to reach out!