Kyle Chu
CS @ Cornell University
Hi! I'm Kyle, a Computer Science student at Cornell University who loves tackling challenging problems through innovative solutions. I'm passionate about building software that makes a real difference in people's lives, whether it's streamlining workflows or solving everyday problems. I enjoy the challenge of solving complex problems and learning new technologies along the way. With hands-on experience in full-stack development and AI engineering, I'm excited to bring my curiosity, creativity, and collaborative spirit to meaningful projects where I can continue growing as a software engineer.
Education
Bachelor of Science in Computer Science
Activities & Involvement
- Pi Delta Psi Fraternity
- Cornell iGem
- Cornell Claude Builders Club
- Cornell University Sustainable Design
- Cornell Chinese Student Association
- Cornell Korean American Student Association
Relevant Coursework
Technical Experience
Teaching Assistant
Key Achievements
- Served as Teaching Assistant for CS 4670: Introduction to Computer Vision, leading office hours and providing technical guidance to 170+ students
- Assisted with the development of assignments, grading guides, and course materials to enhance curriculum quality and student comprehension
- Partnering with a 20-person course staff to streamline grading workflows and provide architectural feedback on student assignments involving convolutional architectures, semantic segmentation, and Vision-Language models using PyTorch and NumPy
Key Achievements
- Developed a node-based AI workflow builder to enable 15+ AEC companies to automate LLM and agent pipelines to reduce workflow execution times with FastAPI, React, LangChain, and PostgreSQL
- Engineered OCR document parsing and structured data extraction pipelines to cut manual document parsing time by 70% using AWS Textract, Pydantic, and PyMuPDF
- Created submittal/due diligence tools to reduced compliance errors by 63% through automated PDF comparison and annotation using Next.js and AWS S3
Key Achievements
- Launched a web platform to streamline updates and boost alumni engagement, increasing viewership by 200% and raising $4,000 using React.js and TypeScript
- Optimized site performance and SEO to improve traffic by 85% through Next.js and efficient asset management
- Delivered a production-quality web app with CI/CD on Vercel, enabling reliable content updates and consistent UX
Key Achievements
- Designed and implemented the Oncurex website for Cornell iGEM team using Figma for UI/UX design, HTML, CSS, and JavaScript, creating responsive web interfaces
- Developed interactive educational gaming application with custom graphics and animations using JavaScript and p5.js library, implementing game mechanics and user interaction systems for STEM education
- Achieved gold medal and "Best Presentation" nomination at iGEM Jamboree 2024, representing Cornell University in international synthetic biology competition with 75,000+ participants and 460 teams worldwide
Skills
Core Languages
Frontend Development
Backend Development
Cloud & DevOps
Database & Storage
AI & Data Science
Projects
Video Game Reviews Analysis
Computational text analysis project investigating linguistic differences between negative game reviews that recommend versus don't recommend, using NLP and statistical modeling
Technologies
Key Achievements
- Processed 371K Steam reviews using VADER sentiment analysis, filtered to 18,898 balanced negative reviews for classification
- Engineered lexicon-based features and applied LDA topic modeling with 7 topics on 5,000-word vocabulary to extract linguistic patterns
- Identified significant predictors and found that monetization mentions 19% less frequent in recommended reviews (OR: 0.82), gameplay topics 8.4x more likely
FantasyForecast
Full-stack app for weekly NFL fantasy point projections
Technologies
Key Achievements
- Built an end to end ML pipeline with a PostgreSQL feature store and per-position XGBoost models validated with expanding timeseries CV and hyperparameter search
- Processed 12+ years of NFL data with automated ETL, feature selection and engineering in Python with Pandas and NumPy
- Outperformed major public fantasy projection systems by achieving up to 45% within ±3 points accuracy and reducing average prediction error by 25% across 12 years of NFL data.
Syllaparse
A full-stack web platform that automatically extracts lecture schedules and exam dates from PDF syllabi, streamlining academic planning for students.
Technologies
Key Achievements
- Used by over 100 students across multiple universities
- Reduced manual calendar entry time by 92%
- Implemented efficient caching with Redis for faster query performance
Sentiment Text Analysis
A multi-class text classifier that predicts 28 different emotions from text input, built using advanced machine learning techniques.
Technologies
Key Achievements
- Processed 15,000+ text samples to build robust dataset
- Boosted prediction accuracy by 74% using stacked classifiers
- Engineered feature extraction pipeline with lemmatization and stopword removal
CUDorms
A full-stack web application for Cornell students to learn more about the dorms and its amenities on campus with a blog section
Technologies
Key Achievements
- Built full-stack MERN application with real-time data updates
- Implemented state management with React Redux for responsive UI
- Developed JWT authentication for secure user data
Latte Link
A mobile app that allows Cornell students to schedule coffee chats with all Cornell clubs that supports calendar events, messaging, and email notifications.
Technologies
Key Achievements
- Architected scalable backend with SQLite and REST APIs
- Integrated SendGrid API for automated notifications
- Containerized and deployed production backend with Docker and Google Cloud Platform