Skylar Shi

I'm Designer

About Me

👋 Hi, I'm Skylar! A UCLA Computer Science student passionate about pushing the boundaries of technology.

Skylar Shi
Web Developer & AI Engineer

Passionate about Full-Stack, Scripting, and Machine Learning

As a UPE Tutoring Chair member, I thrive on leading review sessions for core CS classes and helping fellow students excel in their coursework. My technical journey spans across various domains:

Full-Stack Development

Building robust web applications from front-end to back-end, utilizing modern frameworks and best practices.

Scripting & Automation

Crafting efficient scripts to automate tasks and enhance productivity in various environments.

Machine Learning

Exploring the frontiers of AI, from computer vision to natural language processing, to solve real-world problems.

Data Analysis

Extracting insights from complex datasets to drive informed decision-making and innovation.

Outside of my technical pursuits, I'm an avid golfer ⛳🏌️‍♂️. As my high school's golf team captain, I learned valuable lessons in leadership and precision that I apply to my coding projects. Whether I'm debugging code or on the green, I'm always striving for that perfect execution!

skylarshi@g.ucla.edu
Los Angeles, USA

Technical Skills

Languages & Frameworks

JavaScript (ES6), Python3, C++23, Java, React, Node.js, Express.js, Bootstrap

Machine Learning & Data Science

Pytorch, Tensorflow, OpenCV, LLMs, Prompt engineering

Databases & Cloud Services

MongoDB, SQL, AWS, Azure, Firebase, Supabase

Tools & Technologies

Git, Raspberry Pi, Jupyter Notebooks, Parallel processing

Projects

Redis from Scratch: In-Memory Database Implementation

Built a custom Redis-like in-memory database system from the ground up:

  • Implemented core Redis data structures and commands using C++
  • Developed custom memory management system for optimal performance
  • Created efficient data persistence mechanisms with AOF and RDB
  • Built TCP server handling concurrent connections with epoll
  • Implemented key expiration with a timing wheel algorithm
  • Designed string encoding optimizations for memory efficiency
  • Added support for basic Redis data types: strings, lists, sets
  • Achieved microsecond-level response times for basic operations

This project demonstrates deep understanding of systems programming, database internals, and low-level optimization techniques.

Learn More

AI Technical Interview Simulator

Developed an intelligent mock interviewing platform using advanced AI:

  • Built AI mock interviewer using LangChain and RAG technology
  • Implemented vector embeddings for semantic context matching
  • Created performance scoring algorithm for solution evaluation
  • Designed point-based recommendation engine for personalized practice
  • Architected Next.js/Firebase platform for scalable deployment
  • Deployed containerized application on AWS ECS with blue-green deployment
  • Achieved 99.9% uptime with load balancing implementation
  • Generated dynamic, personalized improvement reports after each session

This project showcases expertise in AI/ML, cloud architecture, and creating production-ready applications with high reliability.

Learn More

Sports Social Platform

Created a comprehensive sports social networking application:

  • Built responsive React Native app for iOS and Android platforms
  • Implemented real-time event tracking and score updates
  • Integrated social features including user profiles and team following
  • Created live chat functionality for game discussions
  • Developed RESTful API backend with Node.js and Express
  • Implemented push notifications for game alerts and updates
  • Utilized MongoDB for efficient data management and querying
  • Integrated with external APIs for live sports data

This project demonstrates proficiency in mobile development, real-time data handling, and creating engaging social features.

Learn More

AI-Powered Face Recognition Door Lock System

Engineered an advanced, secure facial recognition system for smart access control:

  • Leveraged TensorFlow and Python to develop a high-accuracy facial recognition model
  • Achieved 99.7% accuracy in real-time face detection and recognition
  • Utilized OpenCV for efficient image processing and facial feature extraction
  • Implemented on Raspberry Pi for cost-effective, edge-based computing
  • Developed and trained a custom CNN model optimized for edge device deployment
  • Created an automated dataset generation pipeline for model training
  • Optimized CNN architecture and hyperparameters for minimal latency
  • Integrated the AI model with hardware to create a fully functional smart lock solution

This project showcases expertise in AI/ML development, computer vision, edge computing, and IoT integration, demonstrating the ability to create practical, high-performance AI solutions.

Learn More

Yelp Camp: Full-Stack Campground Review Platform

Engineered a feature-rich web application for campground discovery, sharing, and reviewing:

  • Utilized Node.js and Express.js for robust backend development
  • Implemented MongoDB with Mongoose ORM for efficient data management
  • Designed responsive frontend using Bootstrap 5 and EJS templating
  • Secured user authentication with Passport.js
  • Enhanced security using Helmet.js and express-mongo-sanitize to prevent SQL injections and XSS attacks
  • Integrated Cloudinary for streamlined image management
  • Incorporated MapTiler for geocoding and interactive map displays
  • Developed full CRUD functionality for comprehensive user interaction
  • Optimized user experience with visual and location-based features

This project demonstrates proficiency in full-stack development, database management, API integration, and implementation of robust security measures against common web vulnerabilities.

Learn More

Game Development with Unity

I spearheaded the development of Rocket Landing Simulator, a Unity-based game created collaboratively in a game development lab at UCLA. As the lead programmer on a team of four, I focused on implementing the core mechanics that bring the game to life:

  • Engineered realistic rocket physics and movement systems
  • Developed collision detection and response for varied terrain
  • Designed and implemented the multi-level progression system
  • Orchestrated seamless map-to-map transitions

Working within a Git-managed environment, I collaborated closely with team members who contributed additional features such as fuel management, audio integration, and user interface design. This project showcases my ability to create complex game systems, work effectively in a team, and deliver a polished, playable experience.

Learn More
CataractDetection

CataractDetection: AI-Powered Medical Imaging Analyzer

Developed an advanced Convolutional Neural Network (CNN) for medical image classification:

  • Implemented using TensorFlow for efficient deep learning
  • Custom CNN architecture with multiple convolutional and pooling layers
  • Engineered data loaders for seamless processing of medical imaging datasets
  • Incorporated real-time performance visualization and analysis tools
  • Achieved a 97.94% accuracy on the test set

Demonstrates expertise in Python, deep learning, computer vision, and creating AI solutions for healthcare applications.

Learn More

C++ OOP Game Development

Developed a C++ classic game, Marble Madness, showcasing advanced programming skills:

  • Implemented complex game logic and AI behaviors using object-oriented programming principles
  • Designed an extensible class hierarchy for diverse game objects and enemies
  • Created a robust game engine managing real-time user input, object interactions, and level progression
  • Utilized polymorphism and memory management for efficient, scalable code
  • Integrated with existing graphical frameworks, demonstrating adaptability to established systems

This project highlights proficiency in C++, problem-solving, and the ability to work with detailed specifications to deliver a fully-functional, multi-level game experience.

Learn More

C++ UCLA Turn by turn navigation

Developed a sophisticated C++ application to generate optimized walking tours of UCLA and Westwood:

  • Implemented a custom HashMap class for efficient geospatial data indexing
  • Created a GeoDatabase class to load and query Open Street Map data
  • Designed a Router class using A* algorithm for optimal pathfinding
  • Built a TourGenerator to produce detailed, step-by-step tour instructions
  • Utilized object-oriented design principles and STL containers for scalable, maintainable code
  • Optimized algorithms to handle large datasets efficiently (20,000+ street segments)

This project showcases advanced C++ skills, algorithm design, and the ability to work with complex specifications to deliver a fully functional, real-world application.

Learn More

React + Vite Todo App

Engineered a sleek, high-performance Todo application leveraging cutting-edge web technologies:

  • Utilized React for building a dynamic and responsive user interface
  • Implemented Vite as a next-generation frontend tooling for lightning-fast development and optimized production builds
  • Integrated Material-UI components for a polished, professional look and enhanced user experience
  • Developed a local storage solution for seamless data persistence, ensuring user data integrity across sessions
  • Engineered a modular component architecture for improved maintainability and scalability
  • Utilized modern JavaScript features and React hooks for efficient state management

This project demonstrates proficiency in modern frontend development practices, state-of-the-art tooling, and the ability to create intuitive, user-friendly applications. It showcases skills in React, state management, component design, and responsive web development - all crucial for today's fast-paced tech industry.

Learn More

PersonaInsight: AI-Driven Personality Profiling Platform

Developed an interactive machine learning application for advanced personality analysis:

  • Built with Streamlit, scikit-learn, and Agglomerative Clustering
  • Implements Big Five personality model for comprehensive profiling
  • Features real-time survey analysis and global user segmentation
  • Provides instant, AI-generated personality insights and cluster characteristics

Demonstrates expertise in Python, ML, data analysis, and creating user-friendly AI tools for HR and marketing applications.

Learn More

Professional Experience

Machine Learning Intern

Jun. 2024 -- Sep. 2024

Concentric AI, San Jose, CA

  • • Engineered Python automation scripts to implement robust pipelines for Gen AI models, leveraging parallel processing to query Large Language Models (LLMs) concurrently, accelerating OS-level tasks and evaluation processes that would have been 200+ hours to just under 4 hours
  • • Leveraged prompt engineering techniques to optimize AI model performance, resulting in repeatable parsable results
  • • Integrated LLM-context-driven sensitive schema detection solution into internal and production environments, enhancing data security protocols and improving compliance with industry regulations
  • • Conducted benchmarking of multiple lightweight LLMs (8 billion parameters), selecting the top-performing model with a 5% improvement in task-specific accuracy over the average

Software Engineering Intern

Jun. 2023 -- Sep. 2023

Curacao, Los Angeles, CA

  • • Simplified website navigation improving efficiency by 20%, and utilized vendor product optimization, increasing search accuracy by 7%
  • • Researched and proposed 5 alternative SEO strategies to management, one of which had a predicted 30% increase in organic traffic over 6 months

Software Engineering Intern

Aug. 2022 -- Sep. 2022

Curacao, Los Angeles, CA

  • • Developed multiple responsive e-commerce frontend pages for 100+ open-box products
  • • Conducted comprehensive UI/UX analysis by reviewing over 400 User Journey videos, identifying 8 key trends which spurred a solution that led to an 8% improvement in user satisfaction

Information Technology Intern

Jun. 2022 -- Aug. 2022

M-Theory Group, Los Angeles, CA

  • • Led an inventory software migration, collaborating with engineers to implement solutions that significantly improved inventory management processes
  • • Conducted data cleansing of 3,000+ records and delivered training sessions for multiple team members, resulting in 40% enhanced operational capabilities
  • • Developed centralized SharePoint repository for 15 critical WASP (Inventory software) functions, streamlining knowledge transfer and reducing document search time by 30%