Aircraft Maintenance or "Main-tedious"?
Avident to the rescue 👇




Transforming Aircraft Maintenance for Singapore Airlines (SIA) through Computer Vision and Machine Learning
Project Overview
Avident is a cloud-based integrated aircraft maintenance platform for intuitive and effective data management, maintenance execution and reducing oversight. Avident consists of AviHelp and AviManage. AviHelp is the interface for on-the-ground maintenance crew, while AviManage is the interface for executive engineers.
My Role | Lead UI/UX Designer
My responsibilities were mainly spearheading the product design, developing proof-of-concepts for the computer vision and machine learning parts of our solution (using IBM Watson Discovery) and leading the client communications (gathering requirements, iteration feedback, demo pitch). My team consisted of 2 other developers and 1 marketer.
Singapore Airlines (SIA)
"How might we enable our engineers to achieve higher productivity in aircraft maintenance?" - Singapore Airlines’ Business Objective | Full brief
Working demo of AviHelp,
2nd Runner-up of SIA AppChallenge 2019 
Aug 2019 - Oct 2019
Design Tools
Adobe XD, Invision
Tablet App
Strategy & Skills
Requirements gathering, Secondary research, User journey mapping, Wireframing, Rapid prototyping, Personas, User flows, Competitor analysis, UI design

Introducing Avident | AviHelp



For a typical aircraft maintenance task like the anti-skid transducer connector, 73 manuals are needed.

According to the Federal Aviation Administration (FAA), 64% of maintenance accidents are related to challenges with the manuals.

Research Method

Secondary Research


Initially, we had no access to C-suite engineers from SIA, nor on-the-ground maintenance crew. Hence, our team did an extensive requirements gathering through in-depth secondary research to further understand the industry and the problems aircraft maintenance engineers faced. This was especially important for us to understand the context as we were completely new to this field.

How an aircraft maintenance manual typically looks like and how to use it

(Check out the video we made breaking down the problem on the left)

Aircraft maintenance engineers need to cross reference MANY manuals for each maintenance task.

This often causes confusion.

According to the Federal Aviation Administration (FAA), 65% of technicians find the manuals hard to use

This is also extremely (unnecessarily) time consuming.

According to the Federal Aviation Administration (FAA), 50% of total maintenance task time is used to search for information in the manuals.

Because of the sheer amount of information engineers have to reference, they also tend to miss out on warnings in manuals, leading to accidents.

Typically, engineers refer to hard copy manuals. There are existing solutions that try to digitize manuals. However, they remain unintuitive and tedious.

Hard-to read and tedious navigation index

Long lists of chapters without a way to identify the right information and chapter

In A Nutshell

Issue 1: There are just so many manuals to cross reference

This often causes:

confusion (65% of aircraft engineers fine manuals hard to use)
excessive time spent (50% of total task time is used searching for information)

Issue 2: Engineers tend to miss out on warnings because of the sheer amount of information

64% of maintenance accidents are related to challenges with manuals

Issue 3: Existing solutions are still much to be desired

Navigation is still tedious, long lists of chapters without a systematic or easier way to identify the right information



Analysis Methods

Persona, User Journey Map


Using the secondary research data, we developed a persona and user journey map to keep our subsequent designs focused. By synthesizing the data, we were able to deepen our empathy for the engineers and the problem’s context.

Before I get into the nitty gritty parts, I would like to first point out that our solution Avident has 2 parts:

  1. AviHelp for on-the-ground maintenance crew
  2. AviManage for executive engineers

At this current stage of our process, we developed a persona and user journey for on-the-ground maintenance crew (as you can see on the left).

Later on, we were able to talk to C-suite engineers from SIA, which gave us an even more comprehensive understanding of the problem and a new persona, executive engineers.

As you can see in the persona on the left, what is highlighted expands on the effects of the number of manuals needed to be cross-referenced to carry out maintenance tasks and thus the needs, goals and pain points of Arthur.

In the next chapter, I will talk about how I designed the AviHelp platform in a simple step-by-step and intuitively navigable way to account for these key points.

Another thing to note is that Arthur also wants to be able to work well with his team and other teams. This will definitely aid in smoothening out the process of carrying out aircraft maintenance.

In the next chapter, I will talk about how I designed the certification and task-tracking feature that aims to streamline working efficiently within on-the-ground maintenance teams and across teams.

To put things in layman terms, this is how a typical user journey would be for an on-the-ground maintenance crew.

Steps 2 and 3 were the greatest pain points on-the-ground maintenance crew faced. Hence, we made sure to improve the experience of these steps in our digitised platform by adapting the existing step-by-step system used in manuals that is already familiar to crew members into a simpler, more streamlined and intuitive experience with some advanced technology enhancements (which I will touch on in the next chapter).
Additionally, for steps 1, 4, and 5, we made sure to include them in our designs to not deviate too much from their current workflow and minimize the learning curve. The shift from traditional non-digital platforms to digitized platforms can be a scary and intimidating change so we took that into account when designing our platform.


Unify & Update

Low-Fidelity Prototype

To start off, I proposed an initial low-fi prototype to our client. Here are some key features:

Feature 1: Simple split screen interface with step-by-step task instructions with only information you need


Issue 1: There are just so many manuals to cross reference
Issue 3: Existing solutions are still much to be desired

Feature 2: Color code design to easily highlight warnings in manuals to maintenance crew


Issue 2: Engineers tend to miss out on warnings because of the sheer amount of information

Feature 3: Efficient workflow within and between teams (team schedule, task progress and tracker, certification)


Persona and user journey key insights: Arthur wants to be able to work well with his team and other teams.

Interactive Invision Prototype

Parts Recognition

After presenting the low-fidelity prototype to C-suite engineers from SIA, we took their feedback and insights and went back to the whiteboard to draw up wireframes.

As you can see highlighted in the wireframes on the left, we drew up a list of revised features and changes that were incorporated in our high-fidelity prototype and eventually in our working coded demo.

A major change was the addition of AviManage for executive engineers into our platform, with Avident now including AviHelp and AviManage.

I will be bringing you through key design changes made throughout our iterations which resulted in the final product.

AviHelp: Adding ergonomic design features

Taking into account greasy hands of on-the-ground maintenance crew

  1. Button tapping instead of swiping user behavior.
  2. Voice activation to account for difficulty in typing.
  3. Minimize searching: Instead of search fields, clickable buttons directed user to relevant information where possible.
  4. Digital signature: Instead of fingerprint, Face ID or unique phrase voice activation or button click was used.

AviHelp: Keeping the good

Design features kept from our low-fidelity prototype as they received good feedback

  1. Greying out: Users will know where they left off should they go for lunch or toilet breaks.
  2. Maintenance crew are used to the manuals, no matter how tedious it is to use them. Hence, we kept the design to show the original manuals on the right side to increase familiarity and minimize the learning curve of shifting to a new digitized platform.

AviHelp: Adding "Parts Recognition"

We added a new feature: parts image recognition after discovering more struggles in this additional workflow of part identification.

Current workflow to identify a part

  1. Manually flip through a parts catalogue.
  2. Mentally match the part needed to diagrams drawn in the catalogue.
  3. To order a part, need to write down the part number and call to make an order request.

Current workflow is prone to human error, time consuming and unnecessarily tedious.

Adding AviManage

With greater knowledge on SIA’s workflow, we added a new interface, AviManage, for executive engineers who manage the on-the-ground maintenance teams.

Executive engineers are in charge of scheduling maintenance tasks. They select the aircraft, select tasks, plan the timeline and assigns tasks to on-the-ground maintenance teams. At the end of each task, they should receive a notification from the respective team, review work done and issue certification.

With that in mind, the AviManage interface was designed as shown on the left.

UI Design

As our client is SIA, we adapted their brand colors into our high-fidelity prototype.

Client Brand Colors



Interactive prototype

Viv's Reflections & Ramblings

In a perfect world...

We could not conduct contextual inquiry as had no direct access to neither on-the-ground maintenance crew nor executive engineers. Hence, we drew user needs and feedback on features and designs from C-suite engineers from Singapore Airlines Engineering Company (SIAEC). The timeline of 1 month to design and develop a working demo solution was also tight. With more time and direct access to stakeholders, I believe we could have more elegantly designed Avident.

What I've learn...

Understanding something so technical like aircraft maintenance is not impossible. As product designers, it is really our job to try to understand the contexts of our stakeholders to the best of our abilities no matter how unfamiliar the context is, so that we can design the best user experience possible. Although intimidating at the start, I'm glad we pushed through and with a lot of effort and determination, we were able to design a useful product for aircraft maintenance that received a good amount of praise from our client.

Final Favorites

Ending off with my favorite moments!


My favourite moment was when I managed to successfully develop a proof-of-concept for our computer vision parts recognition. As the other developers were busy with other components of our product, it was all hands on board, and I decided to jump in to figure out this feature although I may not have the most expertise in it. Nonetheless, when the image recognition was able to differentiate the parts using my training model, no words could describe the joy I felt. :)

Terrified, scary stage, BUT I DID IT!
Grateful to have the chance to present and pitch to our client in such an arena!