Global Asylum

The mainframe immigration system we replaced, RAPS. A black screen with a mess of different colored text and inputs scattered about.

Mainframe Modernization

It's 2018 and I'm working at United States Citizenship and Immigration Services (USCIS). With over 300,000 cases pending, asylum seekers spend years waiting for a decision. Cases stall in legal or procedural limbo while many applicants wait in detention. Officers decide cases under tight deadlines using a mainframe system built in the 80s.

How do we leverage technology to help officers decide asylum cases faster and *try to* uphold fairness and the law? Chuck the legacy system into the cloud. Establish an intuitive deck of case information cards. Develop flexible design patterns backed by user research. Write new APIs to automate away drudge work, so officers can focus on the applicants.

I was responsible for design, testing, and managing our team's work. Our goal was to cut operating costs and improve security by replacing the old system within a year. Our work saved $10M/year, cut security check turnaround from days to minutes, and reduced manual data entry by 70%.

A refugee officer preparing documents for an interview.

Our persona for an adjudicating officer explaining their story, paint points, and goals. — Our persona for an adjudicating officer explaining their story, pain points, and goals.

Hotline Bling meme of Drake rejecting a single column form in favor of a multi-column form.

Best Practices Not Always Best

One of our usability tests involved a web form using industry best practices: single-column, many pages. This layout usually makes filling out forms less overwhelming. Turns out this pattern doesn’t work when officers need to review large amounts of information at a glance. We opted for a familiar layout mirroring the paper forms that applicants mail to USCIS.

Since officers needed to see all the information at once to make decisions. We designed a deck of cards that grouped related information together. This design made it easier for officers to find the information they needed and make decisions faster.

Hypothesis: Single-column forms reduce cognitive load.

Finding: Officers need to compare information across fields.

Decision: Use card groups matching paper-form mental model.

Result: Faster review, fewer missed details, less stress.

The updated case editing UI utilizes a card based interface to display applicant case data

The updated workflow automated security checks and organized case information into a deck of cards.

Automating Away Drudgery

By talking to users, we learned that their biggest frustrations involved data entry. You know how infuriating it is having to re-type your resume when applying for a job? Well, people at USCIS had to type in all the information from paper forms asylum applicants mailed in. Then they had to retype this information when submitting applicants for security checks.

Our new system automated data entry where possible. We built APIs to automate security check submissions as well. The new system aggregated check results from over eight different law enforcement databases. It also sped up check responses from a matter of days to a matter of minutes.

CHECK RESULTS: DAYS → MINUTES
[ ░░░░░░░░░░░░░░░░░░░░ ] Before: 5 Days
[ █ ] After: 15 Mins

Cost of Waiting: $40/hr x ~1k officers.

Learn About Our Team

Results

$10M in maintenance fees saved

>80% faster security checks