My interest lies in the intersection of AI, Robotics and Human-Computer Interaction, fueled by a profound curiosity about how technology can transform lives and change the world.
This led me to pursue my final-year research thesis on using natural language to command embodied artificial intelligence, which was ranked top 13 in the ALFRED Challenge (2024).
Outside of work, I enjoy working out, exercisng and being a better person than who I was yesterday.
"The struggle you're in today is developing the strength you need for tomorrow." - Robert Tew
Developed, tested, and integrated robotics middleware in healthcare, automating equipment delivery across 3 workflows in 2 hospitals to enhance efficiency and reduce manual labor reliance.
Built TypeScript-based frontend features to enhance UI/UX per client specifications.
Streamlined key workflows by automating Docker builds, code deployment, and log collection with Bash scripts, minimizing manual effort.
Research Goal: Develop an AI model to predict passenger sitting discomfort caused by prolonged sitting in long-haul Business Class flights
using multi-modal sensors and aim to alleviate it using an AI-powered smart seating solution.
Implemented ML algorithms to analyse time-series seat pressure data, studying human shifting patterns over short intervals indicative of
discomfort
Research goal: Enable AI agents to understand high-level natural language instructions and achieve high completion success rate in a suite of simulated scenarios.
Research and incorporate state-of-the-art NLP techniques enabling AI agents to comprehend high level natural language instructions by optimizing LLMs with over 700+ million parameters.
Improved agent decision-making by refining LLM-generated outputs with advanced AI techniques like POMDPs for optimized selection of action sequences.
Achieved top 13th placing on global ALFRED leaderboard outperforming state-of-the-art baseline models.
Developed and implemented computer vision and pattern recognition algorithms to enhance automation products, demonstrating their
capabilities for marketing and client efficiency solutions.
Engaged in meetings with existing and prospective clients to discuss future collaborations.
I have worked on a wide range of projects, from AI/ML research to robotics and even a little bit of software development. Most of my works are school projects or internship related.
Here are some of my notable projects, where I document them in a blog-post format. Click on them to find out more!
Utilize in-context learning (ICL) and prefix tuning to train large language models to identify grammatical error patterns and tendencies in english sentences/essays written by the same author.
Project featured on NUS 24th School of Computing Term Project Showcase (STePS).
[Code][Dataset]
Fine-tuned an LSTM model using self-curated dataset and trained weights to translate customized hand gestures to readable text, in real-time.
A project for a module taken in NUS.
[Code]
Designed a PCB defects tracking system to track 3 types of defects within printed circuit boards, using multi-modal sensors, computer vision and motion control for resolution ehancements.
A project for a module taken in NUS.
[Code]
Skills: CAD • 3D Modelling • SolidWorks • Computer Vision • Python • Circuits • Motion Control • Microcontroller
I took a few courses in AI, and have implemented some paradigms in my assignments and projects. Here are some of the AI algorithms I have worked with:
Reinforcement Learning • Game Theory • POMDP • Q-learning • DQN • A3C • Policy Gradient •
Generative AI • AI Planning and Decision Making • Robotics
Neural Networks
50%
I am comfortable with building simple neural networks and have deployed open source models for larger scale projects. Here are some of the neural networks I have worked with:
Most of my projects are developed in Python, and I have experience with a wide range of libraries and tools. Here are some of the tools and libraries I have worked with:
OpenCV • TensorFlow • Media Pipe • Open Pose • Keras • Sci-Kit Learn • Hugging Face • PyTorch • Pandas • Numpy • Matplotlib • Seaborn •
Plotly • Google Apps Script • Google Apps API
Hardware & Systems
30%
Coming from an engineering background, I offer skill sets in both software and hardware.
GPU • CUDA • Linux • Jupyter • Microcontrollers • Sensors • Servos • Circuits
Software
30%
I use software tools mostly for simulation and CAD for my engineering projects. Never really had the opportunity to hone them.
I am comfortable developing with python, and have used it for most of my projects.
Java
50%
I have taken courses in Java and have used it for development in my school projects. The backend of the General Comfort Rating App was developed in Java.
JavaScript, HTML, CSS
50%
I built this portfolio with JavaScript, HTML and CSS. This project was one of my first few attempts at web development. I have also used javascript in most of UI/UX development in my time at HOPE Technik.
XML
60%
Picked up development using XML for the frontend of the General Comfort Rating App. Safe to say I have gotten the hang of it.
ROS
50%
I have gained extensive experience with ROS while working on robotics middleware framework (RMF) projects at HOPE Technik, where I developed and integrated various robotic solutions to enhance automation in healthcare settings.
C++
20%
Optimised and maintained the RMF backend (written in C++), improving system performance and resolving complex multi-robot coordination issues.
React, Typescript
40%
I have gained experience with React by developing UI/UX components to meet client specifications at HOPE Technik, enhancing the user interface and user experience of various web applications.
Awards
Honoured Graduate NS Specialist Cadet School (2018) - Awarded the Silver Bayonet for top performing cadets.
Edusave CCC-CDC Merit Bursary academic achievement (2016) - Awarded to the top performing students in the cohort.
Dean's List Pre-U 1 (2016) - In recognition of top performing students in the cohort.
MOE Edusave Scholarship (2012 and 2013) - Awarded to top performing students of the cohort.
Model Crestian Award (2013 and 2014) - Awarded to the most outstanding Crestian of the cohort.