A button is placed on every cupholder. Once a user places their cup on the cupholder, the button will be triggered and signals to the MCU that an order was placed. 

An ultrasonic sensor placed right next to the spout could accurately measure fluid level remaining in the cup  and dispense the exact volume of fluid required to top up the drink.

As part of a 2 semester long design project (in a group of 6), I helped to design and prototype a novel crowd management product.

Branded as the FINDR, it helps to do away with the perennial problem of finding tables in crowded hawker centres. Think parking lot indicators, but for food outlets. 

I used Autodesk Fusion 360 to model and render the product, and we utilised C++ with Arduino to develop onboard sensors and light arrays. 3D printed materials (PETG, PLA) were used to construct the main body of the device.

User testing and interviews revealed that our product left positive impressions on both customers and stallowners. 

This competition revolved around streamlining the process of reporting defects from routine MRT track inspections carried out by SBS Transit inspectors. It involved active collaboration with SBS engineers, and numerous interviews carried out with SBS inspectors.

We were able to produce a solution that enabled inspectors to snap a picture with their phone, input in defect details, and send it off to the cloud (AWS S3 and EC2 were utilised), where an engineer back at HQ could analyse defects submitted in real time. I helped to implement machine learning techniques such as the Hough Transform to classify defects into severity levels, for ease of prioritisation by the engineers.

I was in charge of designing the app's UI, and I also designed and built a website portal that could receive data from the app.

The Shell Eco-Marathon is a global competition that challenges engineering students to design, build, and race clean energy vehicles of tomorrow.

I was chiefly responsible for drafting the base sketches from which NTU's next Hydrogen Fuel Cell vehicle (Nanyang Venture 12) would be designed. We then utilised Fusion360 and Solidworks to perform extensive 3D surface modelling to produce the designs for the NV12's carbon fibre-kevlar shell.

I also worked on various improvement projects on NTU's existing vehicles, including refurbishing the NV10's hydraulic braking systems.

I had to adhere to several rules provided by Shell on the design of Hydrogen Fuel Cell-powered vehicles. Other than strict rules that limited dimensions from wheel base and track width, to side mirrow area, I also had to incorporate a sealed bulkhead between the driver and the pressured hydrogen storage tank. The designs above were the basis from where we started to model the vehicle.

During the entire design process of the NV12, we  also worked on the extensive electronic systems on previous vehicles. The NV10 pictured here was also powered by a Hydrogen Fuel Cell. Several Arduino Megas were used to control important onboard sensors and actuators, including controlling the Hydrogen supply. Improvement projects were started to supplement existing designs, such as to improve the software logic of the signalling circuit.

As part of a joint design project between YITU technologies and REP's Makers Club, I revamped the hardware design on the YITU Bot, their early-prototype autonomous patrolling bot, in order to improve its stability and braking performance on inclined surfaces.

I utilised ROS and SLAM and gained some familiarity with Linux terminal commands in order to control bot movement. We designed algorithms to improve the braking logic on the bot once it detected it was on an inclined surface. 

We also implemented hardware changes such as better wheels, a wider base, and an emergency braking mechanism to further improve performance.

Emergency braking systems would help to bypass the issue of insufficient holding torque on the bot's stepper motors. An accelerometer was integrated to trigger the activation of the braking system. 

Features such as a wider base to increase the angle tolerance of the bot as well as better wheels with a higher static friction coefficient were implemented to improve the bot's tolerance to tipping or slipping motion respectively.

This project was for an entrepreneurship course (conducted by University of California Berkeley) that challenged us to create a new tech product in a team of four. 

We developed an idea for a new financial asset platform for retail investors, called AssetPoint. The platform aimed to combine the majority of a users investments (stocks, ETFs, crypto, etc) across multiple platforms into a single unified dashboard. This would enable them to make better informed decisions when data such as whole-portfolio return (%) could be calculated. 

What initially convinced us of possible market demand was the fact that a shocking percentage of investors we interviewed mentioned using excel sheets at some point of their investment journey to summarise information across their investing platforms. 

I was chiefly responsible over product development of the platform and landing page. As part of our user-study phase, I planned and executed several stakeholder interviews to identity early adopters, investigate investor pain points, validate market research, and test out mid- and high-fidelity prototypes.

I spearheaded front-end UI/UX design phase from initial prototype to final deliverable after extensive studying of market trends and stakeholder interviews. I generated extensive wireframes and subsequent high-fidelity prototypes with Figma.

Conducting A/B testing was one of my biggest learning points, testing was conducted amongst potential users to test out and identify most efficient home-page dashboard layouts to improve customer conversion from free to premium model.