Role UI Designer, Software Engineer
Tools Figma, Grafana, Adobe Experience Manager, Jupyter Notebook
Timeline January - June (2020)
Revolutionized server security with user-centric UI dashboard to detect anomalies in server traffic
$2 million
in profits saved by deploying our anomaly detection tool on production servers, ensuring preparedness for malicious traffic.
50+
CMS UI components were developed to optimize functionality and user experience on Air Canada's e-commerce website.
150 hours/ week
were saved for the analytics team through the streamlined anomaly prediction with the tool.
CONTEXT
But, what is a predictive anomaly detection tool?
MOTIVATION
Why did I build this tool?
The tool was developed based on insights from stakeholder interviews and analysis of access logs, indicating the potential of integrating machine learning with real-time access logs to improve anomaly detection. It was tailored for Air Canada's e-commerce website, with a focus on meeting the needs of the Site Reliability Engineering and Analytics teams.
DEFINING SCOPE
Building for the Analytics team: Enhancing monitoring and response
Based on findings and organizational needs, a user-friendly UI dashboard was developed. This would ensure timely decision-making and response without requiring the team to understand complex machine learning code.To assist the analytics team in analyzing and detecting potential anomalies. If anomalies were detected, the tool facilitated alerting the necessary teams to prevent potential damages.
FEATURE IDENTIFICATION
Ideation
What all did the tool need? I jotted down the requirements once I understood what were the expectations:
SETTING EXPECTATIONS
What would success look like?
Effectiveness was gauged by accuracy in predicting cyberattacks, downtime reduction, error rate, and user feedback on the UI's usability.
IDEATION & ITERATION
How do we present all this data?
With the limited screen real estate and the necessity for every dashboard element to be crucial, I initiated the process by sketching out the information architecture. Testing it with wireframes helped gauge user reactions and refine the design.
This was the initial structure of the dashboard, which I tested with 7 members of the analytics team and 6 members of the SRE team.
nTH ITERATION
The new and ~improved~ information architecture
I revamped the information architecture with clearer segregations. To achieve this, I organized a card sorting workshop involving team members from both the Analytics Team and SRE. The objective was to understand how they perceived and categorized the features. Following the workshop, I developed a new diagram and refined the dashboard accordingly. This collaborative effort ensured that the architecture not only met the needs of all stakeholders but also enhanced usability and efficiency.
Accuracy in predicting attacks
Successfully identifiying anomaly and assigning ticket to the correct team by the user
But, we failed here
After testing it with 13 users from the SRE and the Analytics team, I concluded that the design needed further segregation. It should have two distinct parts: an Operations section and an Analytics section, as their purposes vary significantly.
Not all teams perform the same tasks, nor should they have equal levels of permissions and access to data. For instance, the Analytics team's data is less time-sensitive, whereas the Operations team's data is highly time-sensitive, as they are responsible for preemptive tasks in response to anomaly predictions by the system.
FINAL PRODUCT
A glimpse into the final product
LEARNINGS
Reflections
Fostering Stakeholder Engagement
Understanding the significance of active involvement from key stakeholders, I instituted a practice of including leadership in design meetings. This proactive approach facilitated better communication and alignment between different teams, reducing duplicated efforts and enhancing collaboration throughout the design process.
Empowering Teams through Usability
Beyond creating cutting-edge backend technologies, I emphasized the importance of transforming them into user-friendly products that empower teams. By prioritizing usability and intuitiveness in product design, I aimed to equip teams with tools they can easily understand and leverage to drive productivity and innovation.
