I specialize in designing enterprise tools and services that help companies
create and collaborate,
grow and scale,
learn and improve
Work Portfolio
Help.lyft.com
The vault that hosted all internal and external content, self-help tools, and support portals
I lead the foundational competitive research, information architecture redesign, interaction design, and responsive UI redesign for the Lyft Help Center that served millions of external customers and internal teams in multiple countries.
Key accomplishments include an improved search function, dynamic issue identification, embedded self-help tools, a smart escalation form, multi-channel contact, localization, and accessibility features.
Support Platform North Star
Migration, scaling, and redesign of Lyft’s support tools and ticketing system
To meet growing demand, the Lyft team had to rethink the support strategy and consolidate dozens of disconnected tools into one single platform to ensure good customer experience, quality, and efficiency.
I was a key contributor to launching Lyft’s first native chat and multi-channel contact platform built on Twilio Flex. Taking insights gathered from BPO sites, I created service blueprints to shine a light on the overlooked gaps and pain points, and motivated teams and leadership to rethink strategies in large initiatives such as concurrent chat and dynamic case management.
The research I conducted on cross-functional teams set the momentum to kill the old reporting structure and enhanced data quality. The contextual modular design patterns I created with our design team contributed to the early thinking and foundational framework of the North Star vision.
Lyft App Simulator
The internal tool to QA personalized settings and features
There were more than 200 attributes that governed how a help page displayed to a user, such as user types, ride types, languages, regional policies, and devices. The legacy engineering tool for previewing changes required code simulation and was not easily accessible to the Product, Design, Content, and QA teams.
The goals for this project were to democratize simulation access, accelerate development velocity, and increase production quality. I sorted thousands of files in the GitHub repo, distilled the design requirements down to half a dozen modular patterns, and differentiated relevant control mechanisms in a given task for intuitive use. This design empowered our team to reach the project goals and increased our confidence and happiness.
Payment: Authorization
Contextual help timed precisely with empathetic and relevant delivery
Authorizations were severely confusing to riders. People often thought they were overcharged for a ride, charged twice, or charged for a ride they didn’t take. In reality, these were bank verifications, not transactions. There were thousands of banks connected to the user accounts that had different transactional processes and timelines, this further complicates the situation and made this issue among the most painful and highest escalated support cases.
I mapped out all the possible ways a payment authorization could occur and designed anticipatory help interactions that clearly explained the nature of authorizations at the moment confusion arose. It resulted in better customer satisfaction and ultimately led to lower support costs and higher earnings per ride.
Core Quality Control Platform
Automating appraisal review with machine intelligence
HomeVision is a company that aims to build a more transparent, efficient, and equitable housing market. This vision starts with helping lenders and Appraiser Management Companies (AMCs) automate the quality control process on appraisals using AI, NLP, and image recognition technologies.
When I started, HomeVision’s proprietary QC platform was in its MVP stage. Over 2 years, I designed and shipped dozens of functionalities such as image recognition, NLP feedback, auto scoring benchmarks, interactive commentary, and dynamic evidence gathering. I also refreshed the layout and revamped the information architecture to enable smoother navigation and reduce screen fatigue.
Value Review
One-stop shop for analyzing property value and market conditions
When there is an appeal against a property value estimate, there needs to be a bird’s eye view for the reviewer or underwriter to grasp the home quality and market conditions. I spent 4 months working with Product, Engineering, and Data Science to design a dynamic review platform called the “Value Review”. It brings all the analytical tools and data visualizations required into one interface. It maps public data, multiple listing services, and inspection findings to a structured view. This enables the reviewers to effectively determine the precise value of the property and mitigate risk.
Rule Library
An advanced rule decision engine that looks deeper
Rules are the backbones of automation. We had more than 800 rules that govern millions of orders. Before this project, the rules' content, logic, and relationships lived locally on spreadsheets.
I PM’ed and designed the first digital version of the Rule Library. We created an advanced search engine, extensive filtering, and individual rule pages that documented rule descriptions, customer-facing content, application logic, live status, and change history. This library served as the foundation for all internal and external products.