Software Projects

Personal Websites from Scratch

Personal Websites from Scratch

  • Deployed a Next.js application on Azure VM, secured with HTTPS through Nginx, Certbot, and Azure DNS configuration.
  • Implemented Nginx reverse proxy with Certbot for directing 8080 port traffic to HTTPS.
  • Incorporated dynamic descriptions with typing and shining animations and developed reusable components.
Motivation: Self-motivated
Date: Apr. 2023
GitHubGitHub
Webserver & File Sharing System from Scratch

Webserver & File Sharing System from Scratch

  • Developed a concurrent web server using multi-threading, equipped with an HTTP/1.1 protocol handler for request parsing and response formulation.
  • Created a race-condition-free file sharing system with Meta and Block Services, supporting simultaneous file synchronization from multiple clients.
  • Incorporated the RAFT protocol to ensure system fault tolerance.
Motivation: UC San Diego CSE224 Course
Date: Jan. 2022
GitHubGitHub
Buffer Manager and B+ Tree Indexing System from Scratch

Buffer Manager and B+ Tree Indexing System from Scratch

  • Implemented a B+ Tree file indexing system for efficient database operations including insertion, deletion, node splits, and range searches.
  • Developed a Buffer Manager utilizing the Clock Algorithm for optimized buffer replacement.
Motivation: UC San Diego CSE132C Course
Date: Apr. 2022
GitHubGitHub
Recommender System Rating Prediction

Recommender System Rating Prediction

  • Designed a latent factor rating prediction model using TensorFlow, securing a top 5% performance ranking in class.
  • Enhanced the model by integrating it with a deep residual network, resulting in a Mean Square Error (MSE) of 0.83 over 500,000 shopping records.
Motivation: UC San Diego CSE258 Course
Date: Dec. 2021
Computer Vision Projects

Computer Vision Projects

  • Developed edge and corner detection algorithms, utilizing Non-maximum Suppression and Hypothesising techniques.
  • Constructed partially and completely bounded camera rectification solutions using epipolar geometry.
  • Designed a SIFT feature matching system leveraging fundamental metrics computed with epipolar geometry and RANSAC.
  • Employed CNN-Dense classifiers in pattern recognition projects, achieving 98% accuracy.
Motivation: UC San Diego CSE252A Course
Date: Dec. 2021
GitHubGitHub