ProjectsProjectsProjects
pic

NUSMods - Timetable Optimiser

  • Integrated a core feature into NUSMods benefitting over 30,000 students
  • Engineered a high-performance scheduling algorithm in Go using Beam Search with a custom scoring heuristic to solve complex multi-constraint optimisation problems in sub-minute latency.
  • Evaluates millions of timetables based on several user-defined constraints
    • pic

    PortfoliOmatic

  • Built a full-stack robo-advisory platform that delivers personalised portfolio recommendations using Modern Portfolio Theory, the Black-Litterman model, and real-time market data
  • Engineered a Python backend with FastAPI and PostgreSQL (Supabase) for secure authentication, risk profiling, portfolio optimization, capital tracking, and daily performance simulation with scheduled rebalancing logic.
    • pic

    AI Chess Engine

  • Created an AI that thinks 5 moves ahead at a reasonable time (~ 5 seconds)
  • Implemented minimax algorithm optimised with alpha beta pruning
  • Implemented zobrist hashing to cache board state for accurate lookups
  • Other optimisations: Move heuristics, positional advantage, etc.
    • pic

    TrackTern

  • NextJS app to streamline tracking and management of internship application, enhancing the functionality beyond traditional Excel Sheets
  • Implemented analytical features to provide users with clear insights into their application progress.
  • Serving more than 300 active users
    • pic

    HackOMania

  • Within 24 hours, with a team of 4, built a Single Source of Truth web page for Government Schemes, Grants and Loans to help SMEs expand and grow their businesses
  • Implemented several key features like Universal Search, Services page and a personalised chatbot
    • pic

    Text Emotion Recognition

  • Developed a neural network model to detect six emotions in text.
  • Enhanced the model with BiLSTM layers for improved contextual understanding and accuracy.
  • Integrated a pre-trained Word2vec model trained on 100 billion words
  • Processed a training dataset of 30,000 tweets and sentences
  • Achieved 88.3% accuracy on test data.

    • Uniplanner

  • Designed a webapp to assist students plan university life
  • Wrote more than 3000 lines of complex vanilla javascript, html and css to create a single page app
  • Submitted for CS50W project

    • FoodCritic

  • Developed a flask app for logging in, adding restaurants reviews and searching for restaurant info
  • Integrated ChatGPT API into search function for recommendations
  • Submitted for CS50 project