Exit
LUKE MANIMALA
// design

Achieving End-To-End Digitalization For North America’s Largest Independent Aircraft Maintenance Provider

context
I lead product design for the greenfield rebuild of an aircraft maintenance application.
tools
Figma, Avion.io, Azure DevOps
timeline
8 months
No items found.

I lead product design for the greenfield rebuild of an aircraft maintenance application.

This app will eliminate $1M of physical paperwork annually and yield a 20% efficiency gain by person-hours by cutting travel time and paperwork errors.

I lead a team of 4 designers, including myself, with a design budget of $700,000 and a $2.2M development budget over an 8-month project timeline.

Learning the domain by building a user story map

The existing aircraft maintenance process included paper work-cards, which serve as a documentation trail of the steps completed while various individuals work on an aircraft.

We built a user story map centered around the work card’s journey from persona to persona throughout the period before the aircraft arrives to when it is released back to the customer. The workflow included maintenance planners who built out the projects, project managers and supervisors who assigned work cards, technicians who completed the work, inspectors who monitored for quality, QC who ensured that technicians completed all documentation, and finally, records who delivered the end product to the customer with their aircraft. Building the backlog with the story map allowed us to convert our research notes into actionable backlog items, helped us distribute our MVP design and development effort across the end-to-end process, and resulted in a multi-disciplinary product team being abreast of the business processes and research findings.

A dark theme view of a user story map including several small user story cards and a journey labeled "Plan the Work"
A snapshot of our product backlogs user story map

Establishing the base information architecture (IA)

We identified a set of common hierarchal principles among the aircraft maintenance, repair, and overhaul operation (MRO).

Our network has six facilities across North America, called divisions, each holding several aircraft or projects. Each project comprises hundreds of work cards under two categories; routine and non-routine. We leveraged this IA pattern across our modules, where users would land into their default facility, drill into a project for any given aircraft on the site grounds, and navigate into any work card by clicking the table row. We also leveraged table filtering options across the project and work-card list, filter buttons for management that would double as informative displays, and a method for filtering down the large data sets within a project, including over 1,000 work-cards.

A UI screen on tablet device, reading In-Work Projects, with a list of projects
The user lands into their default division and can select the projects within the facility
A UI screen on a tablet device with a list of work cards and a number of Filter Cards across the top with large numbers
The user can filter the work cards in the project using the filters in the table or the filter card UI

Using machine vision and language AI to reimagine the life of a Planner

A .gif of a work card page with animation showing key data points being highlighted
Our machine vision algorithm read each individual page in a work package and parsed out key data that would helped identify the start of each work card

The maintenance planner, responsible for building out the project file with the work cards provided by the customer, is accustomed to managing paper work-card stacks that can easily exceed 6,000 pages for a project.

However, using Azure’s machine vision and language AI suite, we automated several days’ worth of paperwork card splitting into a batch process that takes about two minutes to execute. I helped the development team design the work package splitting algorithm that could accept the customer’s PDF document and process it into hundreds of smaller PDFs, automatically matching the documents to their respective work-card in the application. By implementing this uniquely disruptive method of automating tedious paperwork, we’ll win multiple days a week for the planner persona so that they can focus on more impactful tasks within their expertise, such as building and proposing better project plans and budgets.

A screen on a laptop device showing a document with a number of work card pages and data points in a table
The output of the work package splitting algorithm allowed for user intervention if there was an error in the machine vision

Enabling the digital execution of work cards

Once the work cards are in the system, the aircraft technicians and inspectors execute them, which until this project, our hangar teams would notate using physical stamps and pen notation.

A paper work card with pen markings indicated for mechanic and inspector stamps
An example work card document with pen and ink execution markings

By digitalizing the work card itself, we enabled a host of features that were otherwise only possible by physically moving around the hangar, a space that could often exceed the end-to-end length of a football field. Some examples include stamping and documentation on a work card via e-signature, reviewing availability and checking out parts and tools, finding and attaching maintenance manuals and technical documents, and clocking on to the work card from the tablet rather than a physical time-clock machine. This new UI resulted in substantial efficiency gains as any delivery team member could now access relevant information from anywhere in the hangar, including the aircraft itself, instead of constantly traveling to physical paper trays or computers distributed around the facility.

A UI screen on a tablet with a routine work card on the left and time clock tables displaying how much time was applied to the work card on the right
This view configuration allows the technician and inspector to e-sign the work card on the left and maintain awareness of the time budget using the clock history pane on the right
A dark UI vertically oriented tablet with displaying a table of Work Card attachments
This pane allowed for the technician to upload different reference documents needed to execute the work card
A UI screen on a tablet of a non-routine work card on the left and a pane selector dropdown on the right showing different pane options
The pane selector allows this work card view to be configured in a number of useful combinations the implementation team would require

Enabling faster quality assurance and card completion

A work-card moves across a series of paper bin queues through the rigorous aviation quality assurance process in the legacy paper-based process.

I designed and tested a new method of reviewing work cards that would eliminate the need for paper bins and enable immediate transfer from one reviewer to the other.

In doing this, we created a higher level of transparency in the QA and review process that would otherwise have required an employee to physically walk over to inspect a work card.

We expect to eliminate over 10% of wasted time in work-card QA due to eliminating work-card transportation back and forth between crucial quality assurance checkpoints.

A UI screen on a tablet showing non-routine work card pane on the left and an inspector response modal on the right
The inspector can render a response to the work completed by a technician
A UI screen on a tablet showing a routine work card and a card closure dialog
Production control, or QC, can move the card along through the final quality check process before completion

Disrupting a generations-old way of working

With this product, we aimed to disrupt an MRO operation and culture that has primarily worked the same for the past 50 years.

My goal is to create the most disruptive and humane products possible. If you have a similar mission as me or have any questions about this project, please drop me a line at lukemanimala@gmail.com.

I hope to talk to you soon!

No items found.
No items found.
No items found.
No items found.
No items found.

More case studies